The Biggest Super Bowl Upsets in NFL History, Ranked

Sixty Super Bowls have been played, and the betting market has been wrong about a great many of them. Usually it is wrong by a field goal. Occasionally it is wrong by three touchdowns, and those are the afternoons that get their own documentaries. The New York Jets were 18-point underdogs and won by nine. The New England Patriots were 14-point underdogs and won on the last play of the game. The 2007 Patriots arrived at 16-0, left at 18-1, and have never been able to shake the second number. What follows is the list of the games where the number in the sportsbook window was not merely wrong but embarrassing, ranked by how far the result sat from what everybody expected.

Stylized illustration for The Biggest Super Bowl Upsets in NFL History

How a Real Upset Is Measured

Two numbers matter, and they do not always agree.

The first is the point spread, which is the cleanest statement of what informed people thought before kickoff. A 14-point favorite losing is a bigger shock than a 3-point favorite losing, regardless of how the game felt while it was happening. The second is what was attached to the loss - whether the favorite was merely good or was chasing something historic, because a defeat that ends a perfect season carries weight that a normal one does not.

The games below score high on at least one of those and usually both. Where the two disagree, the ranking leans on the spread, because it is the only part of an upset that was measurable before anybody knew the answer.

Super Bowl III - the 18-point underdog who called it

January 12, 1969. New York Jets 16, Baltimore Colts 7.

The Jets were 18-point underdogs, which remains the largest spread any Super Bowl winner has ever overcome. The Colts were the NFL champions and started Earl Morrall, who had just been voted the league's Most Valuable Player on 33 of 46 votes. The American Football League had lost the first two of these games by a combined 43 points, and the consensus was not that the Jets would lose but that the entire AFL was a lesser competition being humoured with an exhibition.

Joe Namath said publicly, days before, that his team would win. Then his team won, holding the reigning MVP's offense to a single late touchdown.

It is the most consequential upset in the sport's history, because it is the game that made the merger credible. Without it, the AFL is a curiosity that got absorbed; with it, the two leagues were demonstrably playing the same sport. A great deal of what the modern league is sits on top of that afternoon in the Orange Bowl.

Super Bowl XXXVI - the biggest upset since the merger

February 3, 2002. New England Patriots 20, St. Louis Rams 17.

The Rams were 14-point favorites, and at the time the number was not considered unreasonable. They were the most prolific offense of their era and had won the title two years earlier. New England were starting a second-year quarterback who had taken the job after an injury to the incumbent, and the sensible expectation was a long afternoon.

New England won on a field goal as time expired. Tom Brady was named the game's Most Valuable Player, which turned out to be the first sentence of a very long story.

As a pure betting result it is the biggest Super Bowl upset since the AFL-NFL merger, behind only Super Bowl III across the whole series. As a historical hinge it is close to Super Bowl III's equal, because it is the game that started the most sustained run of success any franchise has managed in the modern era.

Super Bowl XLII - the perfect season that was not

February 3, 2008. New York Giants 17, New England Patriots 14.

The Patriots were 12-point favorites and 16-0, the first team to complete a perfect regular season since the schedule expanded to sixteen games in 1978. They were four quarters from the only 19-0 season in the sport's history.

They scored fourteen points. The Giants, a wild-card team that had won three road playoff games to get there, held the highest-scoring offense the league had ever seen to its lowest output of the year and won on a touchdown with 35 seconds left.

What makes this the most painful entry on the list rather than the largest is the asymmetry of what was lost. New England did not merely lose a game, they lost the only version of a record that cannot be shared, and the number 18-1 has followed that roster around ever since.

Editorial illustration: The Biggest Super Bowl Upsets in NFL History

Super Bowl IV - the receipt

January 11, 1970. Kansas City Chiefs 23, Minnesota Vikings 7.

The Vikings were 13.5-point favorites, and the reasoning was that Super Bowl III had been a fluke. The Chiefs had been beaten 35-10 by Green Bay in the first of these games three years earlier, the AFL's overall record against the NFL stood at 1-2, and the prevailing view was that one Jets afternoon did not overturn a decade of assumed hierarchy.

Kansas City won by 16 and Len Dawson was named the game's Most Valuable Player. The final AFL-NFL series finished 2-2, which is the tidiest possible answer to the question of whether the two leagues were equal.

Super Bowl III gets the credit because it came first. Super Bowl IV is the one that made it impossible to argue with.

Super Bowl XXXII - the double-digit favorite falls

January 25, 1998. Denver Broncos 31, Green Bay Packers 24.

The Packers were the defending champions, had gone 13-3, and were favored by double digits. The NFC had won thirteen consecutive Super Bowls at that point - every one of them from January 1985 through January 1997 - and the conference gap was being treated as a structural fact rather than as a streak.

Denver won by a touchdown, becoming the first team favored by ten or more points to lose a Super Bowl since the Vikings in Super Bowl IV. Twenty-eight years separated those two results, which is a useful measure of how rarely the market is that wrong about a championship game.

The streak ended, and the following year Denver won again. What had looked like a permanent imbalance turned out to be a coincidence with a long tail.

Super Bowl XXV - one kick

January 27, 1991. New York Giants 20, Buffalo Bills 19.

The Bills were favored by seven, and the spread understated the gap in how the two teams had arrived. Buffalo's offense was the most explosive in football. The Giants got there with a backup quarterback and a game plan built entirely around keeping that offense on the bench, which they did for more than forty minutes of possession time.

It came down to a field goal attempt in the closing seconds that went wide right. Before the kick, the television broadcast put up a graphic showing that the Buffalo kicker had made one of five career attempts from 40 yards or longer on grass. It is one of the very few times a pregame statistic has predicted a Super Bowl's final play.

Buffalo lost the next three Super Bowls as well. This is the only one of the four where a single play would have changed everything.

Super Bowl LII - the backup

February 4, 2018. Philadelphia Eagles 41, New England Patriots 33.

New England were 5.5-point favorites, and the spread was generous given that Philadelphia had lost their starting quarterback in December and were playing a career backup. The Eagles had been underdogs in all three of their playoff games despite holding the conference's top seed, which is close to an insult in betting terms.

They scored 41 points on a Belichick defense, converted a fourth down on a trick play in which their quarterback caught a touchdown pass rather than threw one, and won the franchise's first Super Bowl. Nick Foles was named the game's Most Valuable Player.

The spread was smaller than the games above it. The result was arguably stranger than any of them.

Super Bowl XLVI - the 9-7 champion

February 5, 2012. New York Giants 21, New England Patriots 17.

The same two teams, four years later, with the same result and one extra wrinkle: the Giants had gone 9-7 in the regular season, the worst record any Super Bowl champion has ever carried into the game. Their win percentage of .563 is the floor for the entire history of the event, and it has stood since.

New England were favored again and lost again by four points. Whatever the market thought it had learned in 2008, it had not applied it.

Honorable Mentions

Super Bowl XL - Seattle were not underdogs by much, and the game is remembered for the officiating rather than for the result, which keeps it off the list proper.

Super Bowl XXXIV - the Rams were favored and won, so it is not an upset at all. It belongs in any conversation about the closest finishes, ending with a tackle one yard short of the goal line on the final play.

Super Bowl XLIX - New England were narrow favorites, so the goal-line interception that decided it was a shock in sequence rather than in outcome.

Closing illustration for The Biggest Super Bowl Upsets in NFL History

Why the Favorite Keeps Losing

Two structural reasons, and neither of them is luck.

The first is that a single game is a small sample of a sport with enormous variance. Over a season the better team wins the vast majority of the time, which is why records separate cleanly by November. Over sixty minutes, one turnover, one missed kick or one injury to a quarterback moves the result more than a fourteen-point talent gap does. Every entry above turned on two or three plays.

The second is that the favorite is usually the team with the more elaborate plan. The Rams in Super Bowl XXXVI, the Patriots in XLII and the Packers in XXXII all arrived with offenses that had a specific way of generating points, and in each case the underdog spent two weeks building a defense designed to take that specific thing away. A team with one great idea is easier to prepare for than a team with three good ones, and the spread does not price that.

There is a third, smaller reason. The two-week gap before a Super Bowl is unique in the sport, because no other game gives a coaching staff that long to design something new. The underdog tends to use the time better, for the simple reason that the favorite mostly wants to keep doing what already worked.

What the Upsets Have in Common

Look at the list and the same shape appears four or five times.

Every winner above generated its pressure from the front without help. The Giants twice beat the Patriots by hitting the quarterback with four rushers, which let seven defenders drop into coverage. New England beat the Rams the same way. Denver beat Green Bay by controlling the ball and shortening the game to fewer possessions than a superior offense needed. None of these underdogs won by out-scheming the favorite in the abstract - they won by making the favorite's best player uncomfortable, then keeping the ball away from him. It is the same principle that puts the greatest pass rushers in NFL history at the centre of every championship defense.

The second commonality is that the underdog's quarterback did not lose the game. Namath, Brady, Eli Manning twice, Foles, Dawson - not one of them produced a huge statistical afternoon, and not one of them made the mistake that decided it. The upset formula is not a heroic quarterback performance, it is an adequate one paired with a defense that wrecks the other team's.

And the third is that the favorites were, in every case, being asked to prove something rather than merely to win. Baltimore was defending the NFL's honor. New England in Super Bowl XLII were defending a perfect season. Green Bay were defending a conference streak. That is the part no spread accounts for, and the list above suggests it is worth a couple of points all by itself.