Flip a Coin

This coin flip gives heads 50% of the time and tails 50% of the time. Each flip is independent: the coin has no memory, so a previous run of heads changes nothing about the next toss. The outcome is decided by crypto.getRandomValues() before the animation starts, which means how long you wait or where you click cannot affect it.

How to Flip a Coin Online

1. Press Flip. The button under the coin, or the space bar, both work.
2. Read the result. Heads or tails appears on the coin and in the result card, and is announced to screen readers.
3. Flip again as often as you like. Every toss is a fresh 50/50 draw.
4. Check the running count. The Results list keeps every flip in order and the tally shows how the split is going. Export CSV saves it.

Nothing is stored on a server and no account exists. Your flip history lives in the page you are looking at and disappears when you close it.

Is an Online Coin Flip Really 50/50?

Yes, and a digital flip is closer to a true 50/50 than a physical coin is.

The generator behind this page is the browser’s cryptographically secure random source — the same primitive used to create encryption keys, not the weaker Math.random(). Out-of-range draws are discarded rather than folded back in, which removes the small bias that a plain modulo would introduce. The full method, including code you can paste into your own console to test the distribution, is on the randomness methodology page.

A real coin is the one that turns out to be slightly unfair. In a 2023 study, Bartos and colleagues recorded 350,757 physical coin flips and found coins landed on the same side they started from about 50.8% of the time — a small but real bias predicted in 2007 by Diaconis, Holmes and Montgomery, who showed that a tossed coin precesses slightly as it spins and therefore spends marginally more time with its starting face up. A digital flip has no starting face and no wobble, so that bias does not exist here.

Coin Flip Probabilities

Each flip is independent, so the chance of a specific sequence is simply one half multiplied by itself once per flip.

Two things people commonly get wrong follow from that table. First, ten heads in a row is unlikely in advance, but once nine have happened the tenth is still exactly 50/50. Second, any specific sequence of ten flips has the same 1 in 1,024 chance — HHHHHHHHHH is no rarer than HTHTTHTHHT. Only the pattern looks special.

The Gambler’s Fallacy

The gambler’s fallacy is the belief that a run of one outcome makes the other “due”. It is the single most common mistake about coin flips, and it is wrong in a specific way worth naming: the coin has no memory and no mechanism by which it could.

After five heads, the next flip is 50/50. After fifty heads, the next flip is 50/50. What actually happens over a long run is not that tails catches up, but that early streaks get diluted — the proportion drifts toward half while the raw difference between heads and tails can keep growing. Statisticians call that regression toward the mean, and it is not the same thing as the coin correcting itself.

Why Streaks Look Suspicious but Are Not

Long runs are far more common than intuition suggests. In 100 flips of a fair coin, a run of at least six of the same side is more likely than not, and a run of five is close to certain.

This is why a genuinely random tool often gets accused of being broken. A student asked to fake 100 coin flips on paper almost always writes too few long streaks, because streaks feel wrong. If you flip here twenty times and see five heads in a row, that is the tool behaving correctly, not evidence of a fault. The tally under the results list is the honest check: over a few hundred flips the split converges on roughly half.

When a Coin Flip Is the Right Tool

A coin flip is also the standard method in sport. An NFL overtime coin toss, the toss that decides who bats first in cricket, and the change of ends in tennis all use one, precisely because it is fast, public and impossible to argue with.

Using a Coin Flip Fairly in Front of People

A draw that is fair and a draw that looks fair are different things, and the second takes a little care.

• Say which side means what before you flip. “Heads, we go to the cinema” settles the argument. Deciding after the coin lands does not.
• Flip on a shared screen. A flip you ran privately and then reported is a claim, not a demonstration. Put it on the projector or share your screen.
• Agree in advance whether there is a re-flip. Best of three is a legitimate format, but it has to be chosen before the first toss, not after a result somebody disliked.
• Do not use it for anything regulated. Licensed lotteries and gambling need certified, audited random number generators. This is a free web page and does not claim to be one.

Best of three is worth understanding rather than just allowing: it does not make the decision fairer, because both sides still have exactly a 50% chance overall. It only reduces how often a single flip feels decisive, which is a psychological benefit rather than a mathematical one.

What This Tool Does That a Real Coin Cannot

Accessibility and Motion

The coin can be flipped with the space bar and the result is announced through a live region, so the outcome is read out without needing to see the animation. If your device is set to reduce motion, the flip resolves immediately instead of animating, and the odds are unchanged — the result is chosen before any drawing happens either way. Sound effects can be switched off and the choice is remembered on that device.

Privacy

Nothing about your flips leaves your device. There is no account, no server-side record of any result, and no tracking of what you decided. The site could not tell you what you flipped yesterday because it never knew.