The idea
Every event either happens or it does not. So the probability that something happens, plus the probability that it does not, equals 1. Rearranged:
P(event) = 1 − P(not the event)
The “not the event” outcomes are the complement. This matters most for “at least one” questions. The opposite of “at least one six” is “no sixes at all”, which is a single clean case. The original question, by contrast, contains many cases: exactly one six, exactly two, and so on.
Example 1: coin flips
What is the chance of at least one head in 5 flips of a fair coin?
Counting directly means adding the cases for 1, 2, 3, 4 and 5 heads. The complement is far shorter: the only way to get no heads is all tails, which is 1 sequence out of 2^5 = 32. So:
P(at least one head) = 1 − 1/32 = 31/32, about 96.9%.
Example 2: dice
What is the chance of at least one six in 3 rolls of a fair die?
The probability of no six on a single roll is 5/6. Rolls are independent, so the chance of no six in all three is (5/6) × (5/6) × (5/6) = 125/216. Therefore:
P(at least one six) = 1 − 125/216 = 91/216, about 42.1%.
Notice that 3 rolls with a 1/6 chance each does not give 3/6. Adding the individual chances over-counts the cases where several sixes appear at once, which is the problem inclusion-exclusion handles from the other direction. The complement route avoids the issue completely.
With 4 rolls the same method gives 1 − (5/6)^4 = 1 − 625/1296 = 671/1296, about 51.8%. So it takes four rolls before an at-least-one-six becomes more likely than not.
Example 3: cards
You draw 2 cards from a standard deck without replacement. What is the chance that at least one is an ace?
There are 48 non-aces among 52 cards. The chance that the first card is not an ace is 48/52. Given that, the chance the second is also not an ace is 47/51, because one non-ace is gone. So:
P(no aces) = (48/52) × (47/51) = 2256/2652 = 188/221
P(at least one ace) = 1 − 188/221 = 33/221, about 14.9%.
Here the draws are dependent events, so the second factor changes. The structure of the shortcut stays the same.
A repeatable recipe
- Write the question in words and spot “at least one” or “one or more”.
- Define the complement, usually “none”.
- Compute the chance of the complement, multiplying across stages if they are independent, or adjusting for removal if they are not.
- Subtract from 1.
When it helps and when it does not
The shortcut shines when:
- the question has a natural “none” case that is easy to count
- there are many trials, so a direct count would need many separate cases
- the exact number of successes does not matter, only whether there is at least one
It helps less when you need exactly two successes, or at most two, because the complement of those events is not simpler. Then a method such as the binomial distribution or a probability tree is more suitable.
Common mistakes
- Forgetting the opposite is “none”, not “one”. The complement of “at least one” is “zero”, not “exactly one”.
- Treating dependent draws as independent. In the card example, using 48/52 twice would give a slightly different and wrong answer.
- Adding instead of multiplying. The chance of no six on three rolls is a product, not a sum.
- Misreading “at least one” as a certainty. A high chance over many tries is still not a promise. Even with a 96.9% chance across five flips, there is a 1-in-32 chance of seeing no heads.
A note on repeated attempts
The same arithmetic describes repeated attempts at anything with a small chance each time. The chance of at least one success rises with the number of attempts, but each attempt on its own keeps the same odds; earlier failures do not make the next try more likely to succeed. In a game with a house edge, more attempts typically mean a larger expected loss, even though the chance of at least one win climbs. Gambling can be harmful, and anyone struggling should look for a local support service.