Is a coin flip really 50/50?

The short answer

For practical purposes, yes. A coin that is caught in the hand or lands on a table comes up heads close to half the time. Careful studies, however, show a small same side bias: a coin flipped by a person tends to land showing the face that was up when it left the thumb. The effect is around one percentage point, so 51 to 49 rather than 50 to 50.

If you need a flip that is exactly fair, use a digital coin like the one on this page. It picks heads or tails with the browser's cryptographic random number generator before the animation starts, so every flip has the same 50% chance and no flipping technique can influence it.

What physics research found about real coins

In 2007, mathematicians Persi Diaconis, Susan Holmes and Richard Montgomery analyzed the mechanics of a flipped coin. A real coin does not only spin end over end; it also wobbles (precesses), and that wobble means it spends a little more time with the starting face pointing up. Their model predicted that a vigorously flipped coin lands on its starting side about 51% of the time.

A large experiment published in 2023 tested that prediction with hundreds of thousands of flips made by dozens of volunteers. The result was very close: about 50.8% of coins landed on the same side they started on. The size of the bias varied from person to person, which fits the idea that it comes from technique.

Two other findings are worth knowing. First, the coin did not favor heads or tails as such: when the starting side was random, the bias vanished. Second, spinning a coin on its edge on a table is a different game. Spinning coins can be strongly biased because the faces are not perfectly identical in weight distribution, so a spin is a worse way to decide than a toss.

How to make a physical flip fair

• Do not look at the coin before flipping, or let another person place it in your hand face hidden.
• Flip high with plenty of rotation. Short, lazy flips with few turns are easier to control, consciously or not.
• Let the coin land on the floor or a table instead of catching and slapping it on the back of your hand.
• Have the other person make the call while the coin is in the air, not before you see how it sits on your thumb.
• If you suspect a coin or a flipper is biased, use John von Neumann's trick: flip twice. Heads then tails counts as heads, tails then heads counts as tails, and two matching results are ignored and redone. Because both mixed orders are equally likely even with a biased coin, the result is fair.

Streaks, and why they do not mean the coin is broken

People expect random flips to alternate more than they do. In reality, streaks are normal. The chance of 5 heads in a row starting at any given point is 1 in 32 (about 3.1%), but in 100 flips the longest run of the same side is usually around 6 or 7. Seeing a run of 6 tails in a long session is not evidence of a problem.

The gambler's fallacy is the belief that after a run of tails, heads is now "due". A coin has no memory: the next flip is still 50/50 whatever happened before. The opposite mistake, the hot hand belief that a side is on a streak and will keep coming, is just as wrong for a fair coin.

How close to 50% should results be?

The law of large numbers says the share of heads gets closer to 50% as the number of flips grows, but small samples wander a lot. Use the Number of coins field above (up to 10,000 at once) to see it yourself. The statistics panel shows the heads and tails count with percentages.

The table shows the range in which about 95% of sessions with a fair coin will fall. Anything inside that range is ordinary luck.

Is best of 3 fairer than a single flip?

With a fair coin, no. Each person still wins exactly half the time whether you play one flip, best of 3 or best of 5. What best of 3 changes is how the result feels: a single flip ends instantly, while a short series gives the loser the sense of a fair chance to come back. If the coin or the flip has a small bias, a longer series does not cancel it; it slightly amplifies it for the favored side.

If the goal is fairness, focus on the method (unseen starting side, high flip, landing on the floor, or a digital flip) rather than on the number of rounds.

Can a coin land on its edge?

It can, but rarely with a normal toss onto a hard surface. A frequently quoted estimate based on a computer model puts the chance for a US nickel at about 1 in 6,000 flips, and thinner coins land on edge even less often. On a soft surface such as a carpet, the coin can lean against fibers more easily. In a game, simply agree that an edge landing is flipped again.