Random Number Generator

Picking a number between 1 and 100 sounds trivial until you need to do it fairly, a raffle draw, a lottery pick, a randomized control group in an experiment. Humans are bad at this. Ask someone to name a random number and they will lean toward 7, avoid round numbers, and unconsciously repeat patterns from the last few numbers they said. This tool generates numbers using the browser's cryptographic random source instead of a human brain or a weak pseudo-random function, with control over range, count, duplicates, decimals, and exclusions.

What this tool does

Set a minimum and maximum, choose how many numbers you want, and generate. Whole numbers or decimals up to six places, your choice. Turn off duplicates to draw unique numbers the way a raffle or a lottery would, each number can only come out once. Exclude specific values, useful for skipping a number that already won a prior round, or one that is unlucky in a particular context. Sort the results in ascending order if you need them in a predictable reading order rather than the order they were drawn.

How to use it, step by step

1. Enter a minimum and maximum value to define the range.
2. Set how many numbers to generate, up to 1,000 in a single batch.
3. Leave decimal places at 0 for whole numbers, or raise it for decimals with that many digits after the point.
4. Optionally list numbers to exclude, separated by commas, if certain values should never come up.
5. Toggle "Allow duplicate numbers" off if you need each result to be unique, like drawing raffle tickets without pulling the same ticket twice.
6. Click Generate, then Copy all to grab every result at once.

Who this is for

Teachers picking a student at random for a question or an assignment use it instead of relying on a memorized rotation students eventually figure out and game. Researchers assigning participants to control and test groups need an unbiased method, and a documented random process holds up better than a coin flip someone claims to have made fairly. Raffle and giveaway organizers use the no-duplicates mode to draw winning tickets without accidentally picking the same number twice. Developers testing an application with randomized input data, load values, sample user IDs, mock scores, pull numbers from here instead of hardcoding the same three test values every time.

Why computers need help being random

A computer following instructions exactly, which is the entire point of a computer, is a strange place to look for genuine unpredictability. Left to its own devices, a naive random function can fall into repeating patterns, especially older or poorly designed pseudo-random number generators that cycle through the same sequence given the same starting point. This tool avoids that by pulling from the Web Crypto API, the same source browsers use for generating encryption keys, which draws entropy from hardware-level noise the operating system collects rather than a predictable mathematical formula.

This matters more than it sounds for anything with real stakes attached. A raffle drawn with a weak random function could, in theory, be predicted or biased toward certain outcomes by someone who understood how the generator worked. Cryptographic randomness closes that gap, the sequence of numbers produced has no detectable pattern, and knowing the last ten results tells you nothing about the eleventh.

Uniform distribution, and why it is not obvious

A fair random number generator should produce every value in its range with equal probability over enough draws, roll a six-sided die a million times and each face should land close to one-sixth of the time. Writing that correctly is easier to get wrong than it looks. A naive implementation using modulo on a random integer,Math.random() % 6 in a language that supports it, introduces a subtle bias called modulo bias, where numbers near the low end of the range come up very slightly more often than numbers near the high end, because the total pool of raw random values does not divide evenly by the range size.

This tool avoids that by scaling the random value directly into the target range rather than using modulo arithmetic, which keeps the distribution flat regardless of what range you choose. Run this generator as a six-sided die a few hundred thousand times and each face lands within a fraction of a percent of the expected one-sixth, no lopsided bias toward any particular number.

Duplicates, and when they should not happen

Whether repeated numbers are acceptable depends entirely on what you are using them for. Rolling a die twice and getting a 4 both times is completely normal, dice rolls are independent events and repeats are expected. Drawing raffle tickets and pulling the same winning number twice makes no sense, once a ticket wins, it is out of the pool. The no-duplicates toggle handles the second case, sampling without replacement so each generated batch contains no repeats, the same principle a physical raffle drum uses when it does not put a drawn ball back in before the next pull.

Decimal mode always allows duplicates, and that is a deliberate choice, not an oversight. With six decimal places of precision across a wide range, the odds of the same exact decimal value coming up twice are so small that enforcing uniqueness would add complexity for a scenario that almost never occurs in practice.

A worked example: running a fair raffle

Say you are drawing three winners from 200 raffle tickets numbered 1 through 200, and nobody should win twice. Set the minimum to 1, the maximum to 200, the count to 3, and turn off duplicate numbers. Generate, and the three results are guaranteed distinct, the tool will not hand you the same ticket number twice no matter how many times you regenerate. If a previous round already awarded ticket 47 and it should not win again, add 47 to the exclude field and every future draw skips it automatically.

Compare that to the informal version of the same task, writing 200 numbers on paper slips and drawing them from a hat. The result is the same in principle, an unbiased draw with no repeats, but a physical drawing takes real setup time and cannot be re-run instantly if a rule changes partway through. The digital version does the identical job, sampling without replacement, in under a second.

Choosing a range that fits the task

Range size changes what a generator can promise. Drawing 5 unique numbers from a range of 1 to 10 uses half the available pool, and the tool will refuse if you ask for more unique numbers than the range contains, an explicit error rather than a silent failure or an infinite wait. Drawing 5 unique numbers from 1 to 10,000 barely touches the pool, so uniqueness is nearly free. Keeping the range comfortably larger than the count you need avoids running into that boundary at all.

Common mistakes when generating random numbers manually

Rolling a physical die or flipping a coin repeatedly introduces bias most people do not notice, a slightly worn die face, a coin with an uneven weight distribution, a person's unconscious tendency to stop shuffling once a sequence looks "random enough" to their eye. Manually typed sequences suffer the same problem from a different angle, someone asked to type out 20 random digits will avoid repeating a digit twice in a row far more often than genuine randomness would, because a repeat subjectively feels wrong even though a fair generator repeats digits constantly over a long enough sequence. Ask ten different people to write down 20 "random" digits by hand and the results will show recognizably similar patterns to each other, more than genuine chance would produce, because everyone is drawing from the same flawed intuition about what randomness should feel like.

This is the core reason software-generated randomness beats manual methods for anything where fairness needs to hold up to scrutiny. A generator does not have an intuition about what randomness is supposed to look like, it applies the same mathematical process every time, which is exactly what makes the output trustworthy.

A note on true randomness versus this tool

A cryptographically secure pseudo-random number generator, which is what powers this tool, is not the same thing as a true random number generator that draws from an unpredictable physical process like atmospheric noise or radioactive decay. For nearly every practical purpose, the distinction does not matter, cryptographic randomness is computationally indistinguishable from true randomness, meaning no realistic amount of analysis can tell the two apart or predict future output from past results. The exception is highly specialized scientific or cryptographic research that specifically requires a hardware-based true random source, a narrow use case this tool is not built for.

If you need randomness for a different kind of decision entirely, picking a name rather than a number, the fake name generator applies the same idea of unbiased selection to names instead of numbers. If your random numbers are feeding into a URL, a query parameter for a test link or a randomized identifier, the URL encoder/decoder makes sure the value survives being inserted into a link intact.

Frequently asked questions