The difference between European and American roulette comes down to one pocket. A European wheel has 37 pockets: the numbers 1 to 36 plus a single zero. An American wheel has 38: the same 36 numbers plus a zero and a double zero. The payouts on the table are the same on both versions, and that is exactly why the extra pocket matters so much. The casino pays as though there are 36 equally likely outcomes, while the ball can land in 37 or 38 places.
How the two wheels are built
In both versions, 18 of the numbers from 1 to 36 are red and 18 are black. The zero pockets are green, belong to neither colour, and count as losses for the even-money bets that depend on colour, parity or range. That detail is the whole mechanism behind the house advantage. A bet on red wins when one of 18 red pockets comes up. On a European wheel, the probability is 18 in 37. On an American wheel, it is 18 in 38.
The pocket count also changes how other bets are described. A straight-up bet covers a single number. A split covers two adjacent numbers, a street covers three, a corner covers four, and a line covers six. Dozens and columns each cover twelve numbers, while red/black, odd/even and low/high each cover eighteen. None of these groupings include the zeros, so adding a second zero never helps the player and only shrinks the share of the wheel that the bettor can cover.
Gross return versus net profit
Roulette payouts are quoted in a way that can confuse newcomers, so it helps to separate two ideas. Net profit is what you gain beyond your original stake. Gross return is the total handed back to you when a bet wins, which includes your stake plus the winnings.
A straight-up bet pays 35 to 1. That is a net profit figure: stake one unit, win, and you receive 35 units of profit plus your original unit back, for a gross return of 36 units. An even-money bet such as red pays 1 to 1: stake one unit, win, and you receive one unit of profit with your stake returned, for a gross return of 2 units. A bet that loses returns nothing, so the net result is minus one unit and the gross return is zero.
Keeping this straight matters when working out expected value, because the calculation can be done either way as long as it is consistent. Using gross return, you multiply the winning payout by its probability and compare the result to the amount staked. Using net profit, you weigh the profit on a win against the lost stake on a loss. Both approaches reach the same answer, but mixing them, for example treating 35 as the amount returned rather than the profit, produces results that look far too generous.
Expected return on a European wheel
Take a hypothetical one-unit straight-up bet on a European wheel. The chance of winning is 1 in 37. When it wins, the gross return is 36 units. When it loses, the gross return is zero. The expected gross return is therefore 36 divided by 37, or roughly 0.973 units per unit staked. Stated another way, the average outcome is a loss of 1 divided by 37, about 2.70 percent of the stake.
Now check an even-money bet on red. It wins with probability 18 in 37 and returns 2 units gross. The expected gross return is 18 times 2, divided by 37, which is 36 over 37 again. The same figure appears for splits, corners, dozens and columns. Because the payout odds are set from a 36-outcome model and the wheel has 37 outcomes, every standard bet carries the same proportional shortfall. That uniformity is one of roulette’s more unusual features: the type of bet changes the volatility, meaning how often you win and how large each win is, but not the long-run average cost.
Expected return on an American wheel
Repeat the same arithmetic with 38 pockets. A one-unit straight-up bet wins with probability 1 in 38 and returns 36 units gross, so the expected gross return is 36 divided by 38, about 0.947 units. The average loss is 2 divided by 38, close to 5.26 percent of the stake. The red bet gives the same result: 18 winning pockets, 2 units returned, 36 over 38.
Put plainly, adding the double zero roughly doubles the theoretical cost of playing the standard bets. A player staking a hypothetical 100 units in total across many spins would expect to give back about 2.70 units on a European layout and about 5.26 units on an American one. These are long-run averages, not predictions for any short session, where results swing widely on either wheel. A single evening can end well ahead or well behind regardless of which version is on the table.
The five-number basket exception
American layouts historically offer one bet that breaks the uniform pattern. The five-number basket covers 0, 00, 1, 2 and 3, and it pays 6 to 1 net, which is a gross return of 7 units. With 5 winning pockets out of 38, the expected gross return is 5 times 7, divided by 38, or 35 over 38. That works out to about 0.921 units, an average loss of roughly 7.89 percent, which is noticeably worse than the 5.26 percent on every other standard bet on the same wheel.
The reason is arithmetic rather than anything hidden. Five pockets do not divide neatly into the 36-outcome payout model, so the payout cannot be rounded up to a fair-looking figure without the house giving something back. The bet exists on the American layout because it covers both zeros, and it is a useful example of why comparing expected return across bets is better than assuming they are all equivalent. It is also worth noting that this bet is specific to the double-zero wheel; a European layout has no equivalent because it has only one zero.
Why a table label is not a rule guarantee
Many readers learn roulette in a context where the label on the table is treated as the definitive description of the game. In reality, the name printed on the layout or shown in a game lobby is only a label. It tells you what kind of wheel the operator says it is offering, but it does not by itself confirm the full rule set. Variations exist in how zeros are treated on even-money bets, whether any partial-return features apply, how bets are settled, and what the payout table actually states. A game marketed as European could in principle carry different settlement terms from the standard model described here, and a game labelled American might use a modified bet menu.
The practical conclusion is to read the rules and payout information that accompany the specific game rather than relying on its name. Any figures in this article describe the standard textbook versions of each wheel using the arithmetic above, and they are not claims about the return of any particular product or provider. Published return figures from an operator or regulator, where they exist, should be treated as the authoritative source for that game, since they reflect its actual configuration.
Reading the difference sensibly
The takeaway for a curious player is simple. The payouts are identical across the two wheels, the straight-up bet is 35 to 1 net on both, but the underlying probability differs, and that shifts the long-run average cost from about 2.70 percent to about 5.26 percent on the standard bets. The five-number basket is the main outlier on the American layout, and the label on a table is a starting point for checking the rules, not a replacement for them. Understanding the distinction between net profit and gross return makes every one of these calculations easier to follow and helps explain why a single extra pocket has such an outsized effect on the maths.