🪙 Coin Flip

Click the coin to flip

Results

Total0 Heads0 Tails0

No flips yet — give it a go!

Free Online Coin Flip — Instant Heads or Tails

This free coin flip tool gives you an instant, unbiased heads or tails result, with no sign-up, no download, and no cookies required. Click the coin — or press Enter when it is focused — and the result appears immediately.

Unlike a physical coin, a virtual coin flip cannot be biased by grip, spin technique, air resistance, or catching angle. Each flip is determined by Math.random(), a pseudorandom number generator built into every modern browser, producing an independent 50/50 result every time. No result is remembered between flips, and no two flips influence each other.

How to use the multi-flip mode

Enter any number from 1 to 1,000 in the count box beside the Flip button. Single flips play a short 0.5-second animation. Multi-flips play one continuous spin (capped at 1.5 seconds), then reveal all results at once. The running stats — total flips, heads count, and tails count — update the moment the animation ends.

Exporting your results

Every flip is logged in the results panel. Once you have at least one result, the Download button becomes active. You can export your full session history as:

JSON — a structured file containing a summary object (total, heads, tails) and an array of every flip with its number and result. Useful for data analysis or importing into code.
CSV — a spreadsheet-compatible file with one row per flip, plus summary rows at the bottom. Open directly in Excel, Google Sheets, or any data tool.

Results are held in memory only and are lost when you close or refresh the page. Download before you leave if you need to keep the data.

Coin Flip Probability — How the Maths Works

A fair coin has two equally likely outcomes on each flip: heads (H) or tails (T). The probability of each is exactly 0.5, or 50%. This is one of the simplest examples of a Bernoulli trial — an experiment with exactly two possible outcomes and a fixed probability.

Independent events

Each flip is statistically independent of every other flip. This means the result of flip 1 has absolutely no influence on flip 2. A common misconception — sometimes called the Gambler's Fallacy — is that after a long run of heads, tails becomes more likely to "even things out." It does not. The coin has no memory. The probability of tails on the next flip is always exactly 50%, regardless of what came before.

Probability of specific sequences

Because each flip is independent, the probability of any exact sequence of n flips is (0.5)ⁿ. The table below shows probabilities for common numbers of flips:

Flips (n)	Probability of all heads	Probability of all tails	Exact probability
1	50%	50%	1 in 2
2	25%	25%	1 in 4
3	12.5%	12.5%	1 in 8
5	3.125%	3.125%	1 in 32
10	≈ 0.098%	≈ 0.098%	1 in 1,024
20	≈ 0.0001%	≈ 0.0001%	1 in 1,048,576

The Law of Large Numbers

Over many flips, the observed proportion of heads and tails will tend to converge toward 50% each. This is the Law of Large Numbers. Try it yourself — flip 1,000 coins using the multi-flip mode and check the live stats. You will almost always see both counters within a few percentage points of 500. The more flips you do, the closer the ratio gets to the theoretical 50/50.

This does not mean the counts will be exactly equal. Random variation (known as standard deviation) means that in 1,000 flips, a difference of 20 or 30 between heads and tails is completely normal and expected.

Standard deviation in coin flips

For n flips of a fair coin, the standard deviation of the number of heads is √(n × 0.25). For 100 flips, that is 5 — so getting between 45 and 55 heads is within one standard deviation and very common. Getting fewer than 40 or more than 60 heads in 100 flips would be unusual (more than two standard deviations from the mean), occurring roughly 5% of the time.

When to Use a Virtual Coin Toss

A coin flip is one of the oldest and fairest methods of random selection. Here are some common situations where a virtual coin flip is genuinely useful:

Making everyday decisions

Deciding who goes first in a game, sport, or activity
Choosing between two equally appealing restaurants, films, or options
Settling a friendly disagreement fairly and instantly
Deciding who does an undesirable chore
Picking which team bats first, defends, or serves

Education and probability experiments

Demonstrating independent events and the Law of Large Numbers in the classroom
Running coin-flip simulations for statistics homework or coursework
Collecting empirical data to compare against theoretical probability
Illustrating the Gambler's Fallacy to students
Generating random binary sequences for computer science exercises

Games and tabletop RPGs

Replacing a physical coin you do not have to hand
Quick random binary outcomes in tabletop role-playing games
Determining random events in board games, simulations, or puzzles

Practical randomisation

Randomly assigning participants to two groups in a study or exercise
Making a start on a decision you have been putting off (if you feel relief at the result, that tells you something)
Breaking a deadlock when two people cannot agree and time is short

Frequently Asked Questions

Is this coin flip truly random?

Each flip uses Math.random(), which is a pseudorandom number generator (PRNG) seeded by the browser using system entropy such as hardware timings. It is not cryptographically secure, but it is entirely suitable for coin flips — the output is statistically indistinguishable from a fair 50/50 split across large samples, and there is no way to predict or influence the result. For everyday decisions, games, and experiments, this is more than sufficient.

Does the site use cookies or store my data?

This site itself sets no cookies and stores nothing — no local storage, no session storage, no tracking. Your flip history exists only in the browser's memory for the current session and disappears when you close or refresh the page. Google AdSense (the ads on this page) may set its own cookies to personalise advertisements. See Google's Privacy Policy for details.

What is the maximum number of flips I can do at once?

The multi-flip mode supports up to 1,000 flips in a single click. All results are calculated instantly; only the animation takes a moment. The results panel displays the most recent 120 flips to keep the page responsive, but your full history is retained in memory and available for download as JSON or CSV.

Why do I keep getting more heads than tails (or vice versa)?

Short runs of apparent imbalance are completely normal. In 10 flips, getting 7 heads and 3 tails is not unusual — it happens roughly 12% of the time. The Law of Large Numbers says the ratio tends toward 50/50 only over many hundreds or thousands of flips. If you are seeing persistent imbalance over hundreds of flips, that would be surprising, but it can still occur by chance.

Can I use this on my phone?

Yes. The site is fully responsive and works on any modern smartphone or tablet. The coin scales with your screen using a fluid size formula, and all controls are touch-friendly. No app download is needed — just open coin-flip.net in your mobile browser.

What format is the JSON export in?

The JSON file contains a top-level object with two keys: summary (an object with total, heads, and tails counts) and flips (an array of objects, each with a flip number and a result string of either "heads" or "tails"). The array is in chronological order, with flip 1 first.

Is a virtual coin flip as fair as a real one?

In practice, a virtual flip is likely fairer. Research has shown that physical coin flips are not perfectly 50/50 — a 2023 study by Bartoš et al. found a slight bias toward the side facing up before the flip, with the same side landing upward about 51% of the time on average. This is because the coin spends marginally more time rotating around one axis. A virtual coin has no physical properties to bias it, so each outcome is equally likely.

Does the light/dark mode preference save between visits?

The site reads your operating system's colour scheme preference on each load using prefers-color-scheme. It does not save a manual toggle choice between sessions (since the site uses no storage). If you switch to light mode and then return later, the site will revert to your OS preference.

Physical Coin Flips vs Virtual Coin Flips

Both methods are accepted as fair in most contexts, but they differ in important ways:

Bias: Physical coins have a documented slight bias toward whichever side starts face up (approximately 51% in favour of the starting side, per the 2023 Bartoš et al. study of 350,757 flips). Virtual flips have no such bias.
Convenience: A virtual flip requires no coin and works instantly from any device with a browser. It is available anywhere, at any time.
Speed: Multi-flip mode lets you perform hundreds of flips in seconds — something impossible to replicate by hand.
Record-keeping: Virtual flips automatically log every result and allow you to export the full history, which is useful for experiments and analysis.
Social acceptance: For in-person decisions where participants can observe a physical coin, some people prefer the tactile tradition of a real toss. Either method is perfectly valid.

Click the die to roll

Results

Total0 Min— Max—

Sum— Avg— Median— Mode—

No rolls yet — give it a go!

Free Online Dice Roller — Roll Any Die Instantly

This free dice roller lets you roll any standard die — d4, d6, d8, d10, d12, d20, or d100 (percentile) — with a single click. No physical dice needed. Every result is produced by Math.random(), giving each face an equal probability on every roll, with no memory of previous rolls.

Choosing your die type

Click one of the die buttons (d4 through d100) to select the die you want. The large die display updates to show the selected type. Switching die type resets the results history so your stats always reflect a single die denomination.

Rolling multiple dice at once

Enter any number from 1 to 1,000 in the count box and click Roll (or click the die). All results are calculated instantly. For a single roll, a short animation plays. For multi-roll, a slightly longer animation runs, then all results appear together. The stats panel — total, min, max, sum, average, median, and mode — update the moment the animation ends.

Exporting your results

Once you have at least one result, the Download button becomes active. Export as JSON (structured data with a summary and per-roll array) or CSV (one row per roll, plus a summary footer).

Dice Probability — Understanding the Numbers

A fair die with n faces produces each value with equal probability of 1/n. A d6 gives each number (1–6) a 1-in-6 chance (~16.7%). A d20 gives each number (1–20) a 1-in-20 chance (5%). The more sides, the flatter the distribution.

Expected value

The expected value (average result) of a single roll of a fair n-sided die is (n + 1) / 2. For common dice:

Die	Faces	Min	Max	Expected value
d4	4	1	4	2.5
d6	6	1	6	3.5
d8	8	1	8	4.5
d10	10	1	10	5.5
d12	12	1	12	6.5
d20	20	1	20	10.5
d100	100	1	100	50.5

Multiple dice and the Central Limit Theorem

Roll a single d6 and any value from 1 to 6 is equally likely. Roll two d6s and sum them — now 7 is the most likely result (1-in-6 chance) because there are more combinations that produce it. As you roll more dice and sum them, the distribution of the total shifts from flat (uniform) toward a bell curve — this is the Central Limit Theorem in action.

Probability of rolling at least n

The probability of rolling a value ≥ k on a dn die is (n − k + 1) / n. For example, on a d20, the chance of rolling 15 or higher is (20 − 15 + 1) / 20 = 6/20 = 30%.

When to Use a Dice Roller

Common reasons to roll a virtual die:

Tabletop RPGs and board games

Dungeons & Dragons and other RPGs — attack rolls, saving throws, damage
Warhammer, Pathfinder, Call of Cthulhu, and other systems
Board games that use non-standard dice (d10, d12, d20)
Quick replacement when a physical die is lost or unavailable

Education and statistics

Demonstrating uniform distributions and expected value in class
Running large-scale simulations (1,000 rolls) to observe the Law of Large Numbers
Comparing theoretical probability against empirical results
Studying the Central Limit Theorem with multi-dice totals

Decision making and games

Assigning tasks or order randomly among a group
Random number generation for puzzles, quizzes, or creative writing prompts
Randomising choices when you have more than two options

Frequently Asked Questions — Dice Roller

Are the dice rolls truly random?

Each roll uses Math.random(), a pseudorandom number generator seeded by the browser's system entropy. The result for a dn die is Math.floor(Math.random() * n) + 1, giving each face an equal 1/n probability. This is more than sufficient for games, decisions, and statistical experiments.

What does d100 mean?

A d100 (also called a percentile die) produces a number from 1 to 100. In tabletop RPGs, it is often used for percentage-based skill checks or random tables. Physically, it is usually represented by rolling two d10s — one for tens and one for units — but a virtual d100 simply picks a single value between 1 and 100 with equal probability.

Does switching die type clear my results?

Yes. When you switch from one die type to another (e.g. d6 to d20), the results list and all statistics are cleared automatically. This is by design — mixing results from different dice would make stats like min, max, and average meaningless.

What stats are shown?

The results panel shows: Total (number of rolls), Min (lowest value rolled), Max (highest), Sum (total of all values), Avg (arithmetic mean, to 2 decimal places), Median (middle value), and Mode (most frequently rolled value). All stats update live after each roll.

What format is the JSON export in?

The JSON file contains a summary object (die type, total rolls, min, max, sum, average) and a rolls array. Each entry in the array has a roll number and a value. The array is in chronological order, roll 1 first.

Can I roll multiple different dice types in one session?

Not in a single results panel — switching die type clears the history. If you need results from multiple die types, roll each type, download the results, then switch. You can then combine the CSV or JSON files for analysis.