What is the most outnumbered victory in history?
In our ranking it is Rorke's Drift in 1879, where about 150 British and colonial soldiers held a mission station against as many as 4,000 Zulu warriors, about 1 to 27, on figures that are unusually well attested. Cajamarca in 1532 is more extreme on paper, but it was an ambush on a largely unarmed escort.
Numbers usually win battles, but not always. These fifteen victories of the smaller side, on land and at sea, from Marathon in 490 BC to Singapore in 1942, are ranked by score, and the odds count the most; see how we chose and ranked them. Every entry says who gives each side's figures and marks the ratios we worked out ourselves. Stands that ended with the defenders destroyed are in our guide to the greatest last stands, and two national stories have their own pages: American victories against the odds and Polish victories against the odds.1
1 Rorke's Drift, 1879
- Where
- Mission station at Rorke's Drift, Natal, South Africa
- War
- Anglo-Zulu War
- Sides
- B Company, 24th Foot, with Royal Engineers and colonial troops, under Lieutenants John Chard and Gonville Bromhead vs Zulu regiments under Prince Dabulamanzi kaMpande
- Strength
- About 140 to 150 British and colonial troops, many of them sick, vs up to 4,000 Zulu warriors; our ratio about 1 to 27 to 1 to 29
- Result
- Post held; 17 defenders killed, 11 Victoria Crosses
On 22 January 1879 a Zulu army destroyed a British column at Isandlwana. That afternoon the small garrison of the mission station at Rorke's Drift, B Company of the 24th Foot under Lieutenant Gonville Bromhead and a party of Royal Engineers under Lieutenant John Chard, learned that thousands of warriors were coming. With hospital patients they could not move fast, they stayed and walled the post with mealie bags and biscuit boxes. The numbers are unusually firm for a colonial fight: the National Army Museum counts 150 defenders against 4,000 Zulus, World History Encyclopedia 140 soldiers against as many as 4,000, and EBSCO 139 against about 4,000. That is about 27 to 29 attackers for each defender, by our reckoning.2
The attackers came from three regiments that had missed the fighting at Isandlwana, led by Prince Dabulamanzi kaMpande, and crossing into Natal went against King Cetshwayo's orders. From about half past four in the afternoon they stormed the barricades for some twelve hours. They set the hospital on fire and broke in, and the defenders fought to get the patients out. When the Zulus drew off before dawn, 17 defenders were dead, almost every survivor was wounded, and some 400 to 500 Zulus lay around the post. Eleven Victoria Crosses were awarded. The 1964 film Zulu, reasonably accurate, made the defense famous.3
Why it ranks here: It is No. 1 on the most extreme odds that are firmly attested: about 150 men against as many as 4,000, with full marks for odds, certainty and memory. Only consequence holds it back, 2, because in the National Army Museum's words it had little effect on the Zulu War as a whole.4
2 Cajamarca, 1532
- Where
- Main square of Cajamarca, northern highlands of Peru
- War
- Spanish conquest of the Inca Empire
- Sides
- Francisco Pizarro's Spanish expedition vs the Inca emperor Atahualpa
- Strength
- 168 Spaniards (Lockhart's study) vs an army of about 80,000 (modern estimate; Xerez, an eyewitness, guessed over 30,000), of whom 5,000 to 6,000, mostly unarmed, entered the square (Hernando Pizarro, eyewitness); our ratio about 1 to 30 in the square
- Result
- Atahualpa captured; about 2,000 Inca dead by Xerez's count (7,000 in a modern account), no Spaniard killed
In November 1532 Francisco Pizarro led a small Spanish expedition to Cajamarca, where the Inca emperor Atahualpa was resting after winning a civil war against his half-brother Huáscar. The Spaniards are counted man by man: James Lockhart's study of the men of Cajamarca counts 168. Pizarro's secretary Francisco de Xerez thought the Inca camp held upwards of 30,000 men; World History Encyclopedia puts the army at 80,000.5
On the evening of 16 November Atahualpa was carried into the square in a litter. Hernando Pizarro, the commander's brother, wrote a year later that about five or six thousand men came with him, without arms. The Spaniards waited in the halls around it. When Atahualpa threw down the book a Dominican friar held out to him, the guns fired and the horsemen charged. The litter bearers and lords around the emperor were cut down, and Pizarro seized him alive. Xerez says it was over in about half an hour, because the sun had already set, and counts 2,000 dead in the square and on the plain; a modern account gives 7,000, with no Spaniard killed. The armed troops outside made no resistance. Myth vs fact: the odds were real, but this was an ambush on a largely unarmed escort, not a battle with the Inca army.6
Why it ranks here: On paper the odds are the most extreme on the list, and the capture opened the conquest of the Inca empire, so consequence 5. But it was an ambush on a largely unarmed escort, so odds 4, and every Inca figure is a Spanish estimate, so certainty 3; it ties Myeongnyang on 40 and goes ahead on recognition.
3 Myeongnyang, 1597
- Where
- Myeongnyang (Uldolmok) Strait, between Jindo Island and the mainland, South Korea
- War
- Imjin War (second Japanese invasion of Korea)
- Fleets
- Joseon navy under Yi Sun-sin vs the Japanese invasion fleet
- Strength
- 13 Korean warships vs 133 Japanese in Yi's war diary (other accounts about 330); both figures Korean; our ratio about 1 to 10
- Result
- Korean victory; 31 Japanese ships destroyed by the Korean count
When Japan invaded Korea a second time in 1597, the Korean fleet under Won Gyun was all but wiped out at Chilcheollyang; 12 warships escaped. Yi Sun-sin was given back command of the navy, and when King Seonjo judged that it could no longer win, Yi answered with words Koreans still quote: “we still have 12 ships left.” A thirteenth came later. The Japanese fleet was pushing west toward the Yellow Sea to support the army on land, and Yi chose to stop it in the narrow tidal channel off Jindo.7
At dawn on the 16th day of the ninth month the Japanese entered the strait. Yi's war diary, the Nanjung Ilgi, counts 133 enemy ships; other accounts, among them a 1999 KBS documentary, give about 330. Even on Yi's figure that is about ten to one, by our reckoning. Yi held the strait until the current turned against the attackers. A Japanese defector then pointed out the body of the commander Madashi in the water; Yi had the head hung up, and Japanese morale broke. By the Korean count 31 ships were destroyed before the rest withdrew. Myth vs fact: the story of an iron chain stretched across the strait to wreck the Japanese ships rests mainly on a family chronicle published in 1914 and is not in Yi's diary or the royal annals.8
Why it ranks here: Thirteen ships against 133 earns full marks for odds; it is a living national story in Korea, so memory 5; it closed the sea route west, so consequence 4. Certainty 3, since every count of the Japanese fleet is Korean; it ties Cajamarca on 40 but falls behind it on recognition, 3 outside East Asia.
4 Singapore, 1942
- Where
- Singapore Island, then a British colony, at the tip of the Malay Peninsula
- War
- World War II, Malayan campaign
- Sides
- Japanese army under Lieutenant General Tomoyuki Yamashita vs Malaya Command under Lieutenant General Arthur Percival
- Strength
- About 23,000 to 36,000 Japanese vs about 85,000 to 100,000 British, Indian, Australian and local troops (counts differ on who is included); our ratio between about 1 to 2.4 and 1 to 4.3
- Result
- Japanese victory; Percival surrendered on 15 February 1942, about 80,000 to 85,000 men captured
By the end of January 1942 the Japanese had driven the British and Commonwealth armies down the length of Malaya, and on 31 January the last of them crossed to Singapore Island and blew up the causeway. Percival had about 100,000 men to hold the island, by Taylor Downing's count. Yamashita had three divisions, and had turned down a fourth to keep his supply line short. Counts of the attackers run from about 23,000 (Downing) to just over 30,000 (James Byrne) and 36,000 in HistoryNet's summary of the battle, which gives the defenders 85,000. That makes the odds between about 2.4 and 4.3 to 1, by our reckoning.9
On 8 February a first wave of 13,000 Japanese crossed the Johore Strait in the northwest and overran some 3,000 Australian defenders; another 10,000 followed within hours. By 13 February the defenders were penned into a perimeter of 28 miles (45 km) around the city. With less than a day of water left, Percival went to the Ford factory at Bukit Timah, Yamashita's headquarters, and signed the surrender on the evening of 15 February. About 85,000 men capitulated, the largest surrender of British-led troops in history; Churchill called it “the worst disaster and largest capitulation in British history.” The factory is now a National Archives of Singapore war exhibition.10
Why it ranks here: Both armies are well documented, so certainty 5, and it was the largest capitulation in British history, so consequence 4. The counts still swing with who is included, so we score the odds at the cautious end, 3; it ties Klushino on 39 and goes ahead on recognition.
5 Klushino, 1610
- Where
- Klushino, west of Moscow, Russia
- War
- Polish-Muscovite War, in Russia's Time of Troubles
- Sides
- Polish-Lithuanian army of Hetman Stanisław Żółkiewski vs the Russian army of Dmitry Shuysky and foreign mercenaries under Jacob De la Gardie
- Strength
- Poles about 6,700 (Rickard's figures, our sum) or 2,700 to 4,000 (Sikora, doubted by Gawron); Russians 30,000 (Rickard; also Żółkiewski's own letter) or about 15,000 (Sikora), plus 3,300 to 7,000 mercenaries; our ratio about 1 to 5 on either count
- Result
- Russian and mercenary army broken; Tsar Vasily Shuysky deposed on 27 July
In 1610, in Russia's Time of Troubles, King Sigismund III of Poland was besieging Smolensk. A Russian army under Dmitry Shuysky, with foreign mercenaries under Jacob De la Gardie, marched to relieve it, and Hetman Stanisław Żółkiewski went to meet it with an army made almost entirely of horse. How many stood on each side is disputed. J. Rickard counts 5,556 hussars, 679 Cossack and 290 Lithuanian horse, 200 infantry and two small guns, against 30,000 Russians and 5,000 to 7,000 mercenaries. Przemysław Gawron's survey of the evidence for the Wilanów Palace Museum sets out the much lower count of the historian Radosław Sikora, 2,700 to 4,000 Poles against about 15,000 Russian soldiers and some 3,300 mercenaries, but Gawron doubts the Polish figure and cites letters from the hetman's side that point to about 5,000 to 7,000; the round 30,000 Russians goes back to Żółkiewski's own letter to the king. Either way the odds were about five to one, by our reckoning, and the Museum of Polish History likewise writes of an enemy army almost five times larger.11
After a night march the Poles reached the camp on the morning of 4 July, but a stout fence held them up. For hours the hussars charged again and again, some companies eight to ten times by the hussar Samuel Maskiewicz's account. The mercenary horse, firing pistols in the caracole, was swept off, and the foreign infantry came to terms once the two Polish guns came up. On 27 July Tsar Vasily Shuysky was deposed in Moscow, and the boyars accepted Prince Władysław as tsar and let Polish troops into the Kremlin.12
Why it ranks here: Full marks for odds, about five to one on both counts, and the victory brought down a tsar, so consequence 4. But the counts vary so widely that certainty is 3, and the name is little known outside Poland and Russia, so recognition 3; it ties Singapore on 39 and sits below it on recognition.
6 Kircholm, 1605
- Where
- Kircholm on the Daugava near Riga (now Salaspils, Latvia)
- War
- Polish-Swedish War
- Sides
- Lithuanian army of Hetman Jan Karol Chodkiewicz vs the Swedish army of King Charles IX
- Strength
- About 3,600 Polish-Lithuanian troops (2,600 horse) vs about 11,000 Swedes (8,358 foot and 2,500 horse, Olofsson); about 1 to 3
- Result
- Swedish army destroyed; 8,000 to 9,000 Swedes dead (estimate), 100 to 200 Commonwealth losses
In September 1605 King Charles IX of Sweden was besieging Riga, and the Lithuanian hetman Jan Karol Chodkiewicz marched to relieve it with about 3,600 men. The Museum of Polish History gives the Swedish army as 11,000, and Magnus Olofsson counts 8,358 foot and 2,500 horse. Chodkiewicz therefore faced about three times his own numbers. In horse the two sides were almost even, 2,600 against 2,500, and his horse had the winged hussars at its core; it was the Swedish infantry that made the gap.13
On 27 September the Swedes held strong ground, and for hours neither side moved. Then Chodkiewicz had part of his force fake a flight. The Swedish line loosened and came forward, and the hussars charged into it. The Swedish cavalry was driven off; the infantry, left on its own, was cut to pieces, and in about three hours the army was destroyed. The Museum of Polish History puts the Swedish dead at 8,000 to 9,000, some 70 percent of the army, against 100 to 200 Commonwealth losses, only 13 of them hussars. The siege of Riga was broken, but the victory brought no strategic gain: unpaid soldiers left the hetman soon after, and the war paused only in a truce.14
Why it ranks here: Odds of about three to one or more, closely counted, so certainty 5, and one of the most one-sided routs on the list. It saved Riga for a time rather than ending the war, so consequence 3, and with recognition 3 outside Poland it ends on 38, a point below Singapore and Klushino.
7 Marathon, 490 BC
- Where
- Plain of Marathon, Attica, Greece
- War
- Greco-Persian Wars (Darius I's expedition)
- Sides
- Athens and Plataea vs the Persian expeditionary force of Datis and Artaphernes
- Strength
- About 10,000 Athenians and Plataeans vs 20,000 to 25,000 Persian fighting men (modern estimates); Nepos, writing over four centuries later, claims 100,000 foot and 10,000 horse; our ratio about 1 to 2 to 1 to 2.5
- Result
- Athenian victory; Herodotus counts 6,400 Persian and 192 Athenian dead
In the late summer of 490 BC a Persian fleet sent by Darius I landed an army at the bay of Marathon, northeast of Athens, to punish the city for backing the Ionian revolt. Athens sent out some 9,000 to 10,000 hoplites and Plataea added 1,000; Sparta would not march before the full moon. Herodotus gives the fleet as 600 triremes but no head count for the army. Cornelius Nepos, writing more than four centuries later, says Datis led out 100,000 foot and 10,000 horse, a foe “of tenfold their own number.” Modern historians cut that to 20,000 to 25,000 fighting men (Mark Cartwright) or about 25,000 (Jona Lendering), about two or two and a half to one, by our reckoning.15
After days of waiting the Athenians under Miltiades charged at a run. Herodotus says the Persians thought them mad, “since they saw how few of them there were,” and that they came without horse or archers. The Greek wings broke the Persian flanks and turned in on the center, and the invaders fled to their ships. Herodotus counts 6,400 Persian dead and 192 Athenian; Lendering thinks the Persian figure exaggerated, though the defeat was heavy. Myth vs fact: Nepos's tenfold odds are a later boast; on modern estimates the Athenians were outnumbered, but by two to one, not ten.16
Why it ranks here: Full marks for recognition and memory, and an invasion turned back, so consequence 4, but odds of about two to one score 3, and certainty only 2, since no ancient writer gives a usable count of the Persian army. It ties five other battles on 36 and heads them: recognition 5 like Agincourt, and greater consequence.
8 Agincourt, 1415
- Where
- Near Azincourt, northern France, about 75 km (45 miles) south of Calais
- War
- Hundred Years' War
- Sides
- Henry V of England vs the French army under the Constable, Charles d'Albret
- Strength
- About 8,500 English vs about 12,000 French (Anne Curry, from the records); Britannica's common estimate about 6,000 English vs 20,000 to 30,000 French; our ratio about 1 to 1.4 to 1 to 5
- Result
- English victory; about 6,000 to 7,000 French dead
Henry V invaded Normandy in August 1415 to press his claim to the French crown. The siege of Harfleur took more than five weeks, and disease, death and desertion thinned his army; he was marching for English-held Calais when a French army barred his road near Azincourt on 24 October. The old story has the French five to one. Britannica gives the common estimate of about 6,000 English against 20,000 to 30,000 French, up to 1 to 5 by our reckoning. Anne Curry of the University of Southampton, working from muster rolls and pay records, puts the English at about 8,500 and the French at around 12,000, odds of only about 1 to 1.4 by our reckoning.17
On St Crispin's Day, 25 October, after a night of rain on freshly plowed fields, the French men-at-arms advanced through the mud between two woods into the arrows of the English archers, sheltered behind sharpened stakes, and the dead piled up in front of the English line. Britannica puts the French dead at about 6,000 and the Pas-de-Calais tourist board at 7,000, against about 400 English by Britannica's estimate. Myth vs fact: the five to one of legend does not survive the records, and Curry argues that a French army that size would not have fitted on the field.18
Why it ranks here: Recognition and memory 5, as the most famous underdog name in English, but the odds are disputed, from 1 to 1.4 on Anne Curry's reading of the records to as much as 1 to 5 on the common estimate in Britannica, so odds 3 and certainty 3. It ties Marathon on 36 and on recognition but follows it on consequence, 3, since by 1453 England had lost everything in France except Calais.19
9 Bannockburn, 1314
- Where
- By the Bannock Burn, just south of Stirling, Scotland
- War
- First War of Scottish Independence
- Sides
- Robert the Bruce, King of Scots, vs Edward II of England
- Strength
- 6,000 to 7,000 Scottish spearmen and 500 horse vs 2,000 to 3,000 English heavy cavalry and about 16,000 foot (Historic Environment Scotland); our ratio about 1 to 2.4 to 1 to 2.9
- Result
- Scottish victory; Edward II fled the field
In June 1314 Edward II marched north to relieve Stirling Castle, besieged by Bruce's brother Edward. Historic Environment Scotland puts the English army at 2,000 to 3,000 heavy cavalry and about 16,000 foot, perhaps with up to 2,000 archers among them, and Bruce's at 6,000 to 7,000 spearmen, trained to fight in tight schiltrons, and 500 horse. On those figures the Scots were outnumbered by about 1 to 2.4 to 1 to 2.9, a ratio we have worked out ourselves. Myth vs fact: John Barbour's poem gave Edward 100,000 men and 30,000 dead, figures Historic Environment Scotland calls far too high.20
On 23 June Sir Henry de Bohun charged Bruce in front of the lines, and the king killed him with a single blow of his axe. Next day the schiltrons did something new: instead of standing to receive the cavalry, they advanced, and drove the English into ground too cramped to fight on. The army broke, the governor of Stirling refused to let Edward in, and the king got away by boat. The victory made Bruce's crown secure at home, but England acknowledged Scottish independence only in 1328. A bronze statue of Bruce, made in 1964, stands on the field.21
Why it ranks here: Memory takes full marks, but odds of about 1 to 2.4 to 1 to 2.9 rest on modern estimates (odds 3, certainty 3), and the war ran on for fourteen more years. On 36 points it trails two better-known battles on recognition and goes ahead of Chancellorsville, equally well known, on consequence.
10 Chancellorsville, 1863
- Where
- Spotsylvania County, Virginia, United States
- War
- American Civil War
- Sides
- Robert E. Lee's Army of Northern Virginia vs Joseph Hooker's Army of the Potomac
- Strength
- About 60,000 Confederates vs about 130,000 Union troops (National Park Service); 57,352 vs 97,382 engaged (American Battlefield Trust); our ratio about 1 to 1.7 to 1 to 2.2
- Result
- Confederate victory; Hooker withdrew across the Rappahannock
At the end of April 1863 Hooker moved against Lee's line at Fredericksburg with what the National Park Service calls 130,000 well-equipped men, against Lee's 60,000. The American Battlefield Trust counts the troops engaged at 97,382 Union and 57,352 Confederate. Either way Lee was outnumbered by about 1 to 1.7 to 1 to 2.2, our ratio from those figures, and few battles on this list have numbers as well documented.22
Lee answered by dividing his smaller army, against every rule. On 2 May Stonewall Jackson took about 30,000 men on a march of more than 12 miles around the Union flank, and at about 5 p.m. fell on Hooker's right and rear, driving the line back more than two miles. That night men of a North Carolina regiment mistook Jackson's party for Union cavalry and shot him three times; he died on 10 May. Hooker pulled back across the Rappahannock on 6 May. He lost 17,304 men, Lee 13,460, which the Park Service reckons at 22 percent of his army. The victory persuaded Lee to invade Pennsylvania, and six weeks later he was at Gettysburg.23
Why it ranks here: Well-documented figures make these some of the surest odds on the list (certainty 5), but a gap of 1 to 1.7 to 1 to 2.2 scores 3, and the win led to Gettysburg rather than ending the war, so consequence 3. On 36 points it follows Bannockburn on consequence and leads Okehazama and Plassey on recognition.
11 Okehazama, 1560
- Where
- Okehazama, Owari Province (today Aichi Prefecture), Japan
- War
- Wars of the Sengoku period
- Sides
- Oda Nobunaga vs Imagawa Yoshimoto
- Strength
- 2,000 to 3,000 Oda troops vs about 25,000 Imagawa (the usual figure in Japanese accounts; 45,000 in the chronicle Shinchō kōki); our ratio about 1 to 8 to 1 to 12
- Result
- Oda victory; Imagawa Yoshimoto killed
In the summer of 1560 the warlord Imagawa Yoshimoto marched into Owari, the home province of the young Oda Nobunaga. The usual figure for his army is about 25,000; the Shinchō kōki, the life of Nobunaga written by his retainer Ōta Gyūichi, gives 45,000, which Japanese writers today treat as too high. Nobunaga could raise about 3,000, and when he rode out to the fort at Nakajima he had fewer than 2,000 with him. That makes odds of about 1 to 8 to 1 to 12, our ratio, but every number goes back to chronicles written later, and they are argued over.24
Nobunaga's outnumbered army caught Yoshimoto's column at Okehazama, broke it and killed Yoshimoto. Myth vs fact: the famous version has Nobunaga circle through the hills in a rainstorm and fall on a camp at rest, but the historian Fujimoto Masayuki, reading the Shinchō kōki closely, argued that he made no detour and won with a frontal attack; others find a frontal win by 2,000 men over 25,000 hard to believe, and the debate goes on. What is not in doubt is the result. Yoshimoto was dead, and Nobunaga went on to subdue his rivals and take Kyoto in 1568.25
Why it ranks here: Consequence takes full marks, because it launched the career of the man who began the unification of Japan, and the usual figures give odds of 1 to 8 or worse. But they rest on later chronicles, so odds 4 and certainty 2; on 36 points it is level with Plassey on recognition and consequence and goes first as the older battle.
12 Plassey, 1757
- Where
- Plassey (Palashi), Bengal, India
- War
- East India Company war with the Nawab of Bengal
- Sides
- Robert Clive (East India Company) vs Siraj ud-Daulah, Nawab of Bengal
- Strength
- About 3,000 (some 800 Europeans and 2,100 sepoys, National Army Museum; World History Encyclopedia counts about 700 Europeans and 1,400 sepoys) vs about 50,000 on paper, much of which never fought; our ratio about 1 to 17 on paper
- Result
- Company victory; Clive lost 22 killed and 50 wounded
On 23 June 1757 Robert Clive faced the army of Siraj ud-Daulah outside Plassey with about 3,000 men: some 800 Europeans and 2,100 sepoys, with ten field guns and two small howitzers. The National Army Museum puts the Nawab's army at around 50,000, including 16,000 cavalry, with 50 guns served by French officers; Clive himself wrote of 15,000 horse and 35,000 foot. On paper that is about 1 to 17, by our reckoning. But the paper odds are misleading: Mir Jafar, who commanded the Nawab's cavalry, had already been secretly offered the throne, and on the day he refused to take part.26
A heavy downpour decided the fight. Clive's gunners kept their ammunition dry under tarpaulins, the Nawab's did not, and his guns fell silent; the cavalry he sent in after the rain was cut down, and most of his infantry left the field. Clive counted 22 killed and 50 wounded, and about 500 of the enemy killed. Mir Jafar became Nawab, little more than a puppet, and in 1765 the Company won the right to collect the revenues of Bengal. No source we opened counts how many of the Nawab's men actually fought, so we score the odds below the paper figure, at about 1 to 2 or more; that score is our own judgment.27
Why it ranks here: Consequence takes full marks, since it opened British rule in Bengal, but the paper odds mean little when part of the enemy army had been bought, so odds 3 and recognition 3. On 36 points it is level with Okehazama on recognition and consequence and comes after it as the younger battle.
13 Cannae, 216 BC
- Where
- Plain by the Aufidus river, Apulia, southern Italy
- War
- Second Punic War
- Sides
- Hannibal's Carthaginian army vs the consuls Aemilius Paullus and Terentius Varro
- Strength
- About 40,000 foot and 10,000 horse vs about 80,000 foot and rather more than 6,000 horse (Polybius); our ratio about 1 to 1.7, nearer 1 to 1.5 on the field
- Result
- Roman army destroyed; about 70,000 dead (Polybius)
Cannae is the classic case of the smaller army destroying the larger, and it is true, but the gap was smaller than its fame suggests. Polybius, the main ancient source, gives Rome about 80,000 foot and rather more than 6,000 horse, and Hannibal about 40,000 foot and 10,000 horse. That is about 1 to 1.7, our ratio. Paullus also left 10,000 foot in his camp, so on the field itself the odds were nearer 1 to 1.5.28
Hannibal pushed his Spanish and Celtic infantry forward in a crescent and let it give way under the legions, while his Africans waited on either side; when the Romans pressed into the center, the Africans wheeled inward on their flanks and his cavalry closed the rear. Polybius counts about 70,000 Romans dead and puts Hannibal's losses at about 4,000 Celts, 1,500 Spaniards and Africans and 200 horsemen; these are ancient figures. Tarentum and other towns went over to Hannibal, yet Rome refused to give in and, as Polybius says, went on to defeat Carthage. The name outlived the war: Schlieffen made Cannae the model for his idea of envelopment and annihilation, and in 1914 the younger Moltke told his son to study it.29
Why it ranks here: Recognition and memory take full marks, and Polybius gives unusually firm figures for an ancient battle (certainty 4). But Hannibal was outnumbered by well under 1 to 2, so odds 2, and Rome won the war, so consequence 3; it leads the battles on 35 points by recognition.
14 Narva, 1700
- Where
- Outside the fortress of Narva, Swedish Estonia (today Estonia)
- War
- Great Northern War
- Sides
- Charles XII of Sweden vs the Russian army besieging Narva
- Strength
- 10,537 Swedes (Martinsson) or about 8,000 (HistoryOfWar, EBSCO) vs 35,000 to 38,000 Russians; our ratio about 1 to 3.3 to 1 to 4.8
- Result
- Swedish victory; 8,000 to 9,000 Russians killed and wounded (Martinsson), 10,000 dead in HistoryOfWar
In autumn 1700 some 35,000 Russian troops of Peter the Great laid siege to the Swedish fortress of Narva, and Charles XII marched to its relief. Örjan Martinsson's study of the battle counts the Swedish army at only 10,537 men against 35,000 to 38,000 Russians; HistoryOfWar and EBSCO give Charles about 8,000. Either way he was outnumbered by more than three to one, about 1 to 3.3 to 1 to 4.8 on our arithmetic. Peter himself had left the camp on the eve of the battle.30
On 30 November 1700 (20 November in the Swedish calendar) Charles attacked the long Russian siege lines at once. Heavy snow blew into the Russians' faces and hid the Swedes until they were on them. Martinsson puts the Russian killed and wounded at 8,000 to 9,000, HistoryOfWar gives 10,000 dead; Martinsson counts 667 Swedes killed and 1,247 wounded, our sums of his figures. Charles kept only the officers, ten generals among the prisoners, and let the rank and file go home. It did not last. Peter used the respite to build a modern army, beat Charles at Poltava in 1709, and came out of the war in 1721 as a Baltic and European power.31
Why it ranks here: Odds of more than 1 to 3 on well-kept Swedish figures score 4 for odds and 4 for certainty, but Russia recovered and won the war, so consequence 3, and few outside Sweden and Russia know the name. It ties Cannae and Blood River on 35 points: behind Cannae on recognition, ahead of Blood River.
15 Blood River, 1838
- Where
- By the Ncome River, KwaZulu-Natal, South Africa
- War
- Voortrekker war with the Zulu kingdom
- Sides
- Voortrekker commando under Andries Pretorius vs the army of the Zulu king Dingane under Dambuza (Nzobo) and Ndlela kaSompisi
- Strength
- About 470 Voortrekkers (South African History Online; 464 and three English settlers in George Chadwick's account, with some 340 servants and allies) vs 10,000 to 20,000 Zulu warriors (South African History Online; Chadwick 12,000 to 15,000, some accounts 32,000 to 40,000); our ratio about 1 to 21 to 1 to 43
- Result
- Laager held; more than 3,000 Zulu dead, no Voortrekker killed and three wounded
In February 1838 the Zulu king Dingane had the Voortrekker leader Piet Retief and some 70 of his followers killed, and fighting between the Zulus and the Voortrekkers went on through the year, as George Chadwick tells it in his account for the South African Military History Society. In November Andries Pretorius arrived with a cannon and was elected commandant general, and in December he led a commando into Zulu country. On 15 December it formed a laager of wagons, probably 64, between a deep donga and the Ncome River. South African History Online counts 470 Voortrekkers against 10,000 to 20,000 Zulu warriors; Chadwick thinks 12,000 to 15,000 more likely than the 32,000 or 40,000 of some accounts. That is about 21 to 43 Zulus for each Voortrekker, by our reckoning, and still about 15 to 1 if Chadwick's servants and allies are counted.32
The Zulu regiments attacked at dawn on 16 December. Musket fire and grapeshot from the laager cut them down, and in Chadwick's reading the cannon did great damage; many attackers took cover in the donga, where the Voortrekkers attacked them, and mounted men rode out and pursued the rest. The fight was over by midday. More than 3,000 Zulus fell; no Voortrekker was killed and only three, Pretorius among them, were wounded. The Ncome, red with blood, became Blood River. Survivors built a cairn on the laager site in 1866, a monument followed in 1938, and since 1995 the anniversary, 16 December, has been South Africa's Day of Reconciliation.33
Why it ranks here: Odds of about 1 to 20 or worse on the usual figures earn full marks, but recognition is only 2, since few outside South Africa know the name, and Chadwick warns that few original sources survive, so certainty 3. It ties Cannae and Narva on 35 points and comes last of the three on recognition.
How we chose and ranked these battles won against the odds
We counted a battle when the clearly smaller side won it, on land or at sea, in any country and any age. A post that held and beat off the attack counts, which is why Rorke's Drift is here; a stand that ended in the defenders' destruction, such as Thermopylae or Zadwórze, belongs to our guide to the greatest last stands. Rorke's Drift is also told there, but for this page we leave that guide only the stands that were lost. A win counts only if the winner was clearly the smaller side on modern estimates, not just in the chronicles: Gaugamela, Issus and Vienna are famous as underdog victories, but modern estimates leave their winners only slightly outnumbered, if at all. Wars as a whole, such as the Six-Day War, are left out, and so are famous wins where the winner was in fact the larger army on the day, such as Trenton. The national stories have their own pages: American victories against the odds and Polish victories against the odds.
We considered 28 candidates and scored each one from 1 to 5 on five criteria. The odds count three times, because this page answers the question of which victories were won against the heaviest odds; recognition, consequence and certainty of the numbers count twice, and memory counts once. Ancient and medieval army sizes are often inflated by the chroniclers, so we scored the odds on modern estimates and marked down certainty where the figures are weak:
- Recognition (×2). 5 if the name is known far beyond military history and appears on most published lists of its kind, 1 if only specialists know it.
- The odds (×3). How outnumbered the winner was, on sourced modern figures: 5 if about 1 to 4 or worse and well attested, 4 about 1 to 3, 3 about 1 to 2, 2 a clear but small gap (about 1 to 1.5), 1 if the sides were roughly even or the winner was possibly not smaller; a 1 puts a battle out of scope.
- Consequence (×2). 5 if it decided a war or the fate of a state, 1 if it changed nothing about the outcome.
- Memory (×1). 5 if it is still a living tradition (a day of remembrance, a monument, a song, a national story), 1 if it is forgotten.
- Certainty (×2). 5 if both sides' numbers are confirmed by records or independent sources, 1 if they rest on one late or partisan source. Ancient and medieval odds are often inflated and score low.
The list is ranked by total score, not by date. Ties go to the better-known battle, then to the one with the greater consequence, then to the older one. The scores are our editorial judgment, made on the evidence in the sources below, and the table shows every one of them, so you can see exactly why Rorke’s Drift, about 150 men against as many as 4,000, is first while Cajamarca, more extreme on paper but an ambush with figures that are all Spanish estimates, is second, and why Cannae, perhaps the most famous battle on the list, sits only thirteenth: Hannibal was clearly outnumbered, but by less than two to one.34
| Rank | Name | Recognition ×2 | The odds ×3 | Consequence ×2 | Memory ×1 | Certainty ×2 | Total |
|---|---|---|---|---|---|---|---|
| 1 | Rorke's Drift | 4 | 5 | 2 | 5 | 5 | 42 |
| 2 | Cajamarca | 4 | 4 | 5 | 4 | 3 | 40 |
| 3 | Myeongnyang | 3 | 5 | 4 | 5 | 3 | 40 |
| 4 | Singapore | 4 | 3 | 4 | 4 | 5 | 39 |
| 5 | Klushino | 3 | 5 | 4 | 4 | 3 | 39 |
| 6 | Kircholm | 3 | 4 | 3 | 4 | 5 | 38 |
| 7 | Marathon | 5 | 3 | 4 | 5 | 2 | 36 |
| 8 | Agincourt | 5 | 3 | 3 | 5 | 3 | 36 |
| 9 | Bannockburn | 4 | 3 | 4 | 5 | 3 | 36 |
| 10 | Chancellorsville | 4 | 3 | 3 | 3 | 5 | 36 |
| 11 | Okehazama | 3 | 4 | 5 | 4 | 2 | 36 |
| 12 | Plassey | 3 | 3 | 5 | 3 | 4 | 36 |
| 13 | Cannae | 5 | 2 | 3 | 5 | 4 | 35 |
| 14 | Narva | 3 | 4 | 3 | 3 | 4 | 35 |
| 15 | Blood River | 2 | 5 | 3 | 4 | 3 | 35 |
Close calls
Five battles came close and did not make the fifteen:
Salamis, 480 BC. Famous and decisive, but its odds shrink under scrutiny: Aeschylus gives the Greeks 310 ships and Herodotus 380, while the 1,207 Persian ships of both writers come down to perhaps 500 in a modern estimate, with no consensus. That is no worse than about 1 to 1.6, so odds 2, and certainty 2.35
Carrhae, 53 BC. World History Encyclopedia gives Surena about 10,000 horse archers and 1,000 cataphracts against some 36,000 Romans, odds of about 1 to 3.3 by our reckoning, and the Parthians destroyed Crassus' army. But almost every figure goes back to Plutarch, who gives only Surena's own following of no fewer than 10,000, and Jona Lendering argues that the Parthians were in fact superior in numbers as well as quality. With the odds in doubt, certainty drops to 2 and Carrhae ends on 33 points, just off the list.36
Assaye, 1803. Arthur Wellesley attacked a Maratha army of at least 40,000 men and more than 100 guns with about 6,500 men and 22 guns, and lost a third of his force winning. He later called it the best battle he ever fought, but it is little known today and its memory has faded, which keeps it off the list.37
Aljubarrota, 1385. Castile brought perhaps 31,000 to 44,000 men into Portugal, but the Aljubarrota Foundation reckons only about 10,000 of them fought, against about 7,000 Portuguese, and João Gouveia Monteiro argues the Castilians had no numerical edge in the battle at all. It secured Portugal's independence, but few outside Portugal know the name.38
Crécy, 1346. The longbow's great day is famous, but the World History Encyclopedia puts Edward III's army at perhaps 12,000, or 15,000 in some estimates, against about 25,000 French, odds of about 1 to 1.7 to 1 to 2. That scores 2, and Agincourt tells the same story higher on the list.39
