Random Number Generator

A random number generator produces values that are unpredictable and, ideally, uniformly distributed — meaning every possible number in your chosen range has an equal chance of being picked, with no bias toward any particular value. This matters for anything from picking a raffle winner fairly, to shuffling a set of quiz questions, to seeding a simulation or a game mechanic that shouldn't be predictable or gameable by users. Not all "random" number generation is equally trustworthy. A very common but weaker approach uses a general-purpose pseudo-random source designed for things like animations or simple games, but not intended for anything where unpredictability actually matters, such as raffles, lotteries, security tokens, or anything users might have an incentive to predict or manipulate. This tool instead uses the Web Crypto API's cryptographically secure random source, which pulls from the operating system's own random number source — the same quality of randomness used to generate encryption keys — giving genuinely unpredictable results rather than merely "good enough for a game" ones. Getting a uniform distribution from raw random bytes is trickier than it sounds. If you naively take a random 32-bit number and reduce it to your range using the remainder operator, some numbers in your range end up very slightly more likely than others whenever your range doesn't evenly divide the total number of possible 32-bit values, a subtle bias called "modulo bias". This tool avoids that by using rejection sampling: it repeatedly draws a fresh random 32-bit value and only accepts it if it falls within the largest span that divides evenly by your range size, discarding and re-drawing any value that would introduce bias. The result is a genuinely uniform distribution across your Min-Max range, no matter what the range size is. Worked example: setting Min to 1, Max to 6, and Count to 1 simulates a single fair six-sided die roll — over many repeated presses of Generate, each of the six outcomes should occur roughly equally often, thanks to the rejection-sampling approach described above. The "Unique values only" option is useful whenever repeats would be meaningless or unfair, for example drawing 5 raffle winners from 50 entries, or dealing out a set of distinct random positions. Internally, when this option is checked, the tool builds the complete list of every integer in your range, then shuffles that list using the Fisher-Yates algorithm, a well-studied, unbiased shuffling method that guarantees every possible ordering is equally likely, before taking the first "Count" values from the top. This differs from simply generating random numbers and discarding duplicates one at a time, which becomes slow and can technically bias later selections as the pool of acceptable values shrinks. A common misconception is that requesting, say, 10 unique values from a range of only 5 possible integers (1 to 5) should just return however many unique values it can, silently ignoring the rest of your request. Doing that quietly would hide a real input mistake. Instead, this tool tells you directly how many unique values are actually possible in your chosen range, and caps generation there, so you immediately understand why you didn't get the count you asked for.