Visual Tools
Calculators
Tables
Mathematical Keyboard
Converters
Other Tools

How to Approximate a Binomial with a Poisson

When a binomial has many trials and a small success probability, its counts behave like a Poisson variable with the same mean λ=np\lambda = np. The Poisson formula has no large binomial coefficient and no huge power, so sums stay short.

Recognize the task

A binomial count with a large nn and a small pp, so successes are rare: "10001000 items, each defective with probability 0.0020.002", "500500 policies, claim rate 0.4%0.4\%". Test: n>20n > 20 and p<0.05p < 0.05 (the rule of thumb on the site). If pp is close to 11, count the failures instead.
The link between the two models appears in the Poisson distribution lesson.

Two ways to do it

Poisson with λ = nprare events
(nk)pk(1−p)n−k≈e−np(np)kk!\binom{n}{k} p^k (1-p)^{n-k} \approx \frac{e^{-np}(np)^k}{k!}
Use the Poisson PMF with λ=np\lambda = np; then follow Poisson probability.
Exact binomialsmall n
P(X=k)=(nk)pk(1−p)n−kP(X = k) = \binom{n}{k} p^k (1-p)^{n-k}
With few trials, or with software at hand, compute the binomial PMF directly.

Steps

StepOn n=500n = 500, p=0.004p = 0.004, P(X=3)P(X = 3)
1Check the conditions. The count is binomial, nn is large and pp is small.n=500>20n = 500 > 20, p=0.004<0.05p = 0.004 < 0.05 ✓
2Set λ=np\lambda = np. The Poisson keeps the binomial mean.λ=500⋅0.004=2\lambda = 500 \cdot 0.004 = 2
3Compute with the Poisson PMF. Add terms or use the complement exactly as for a Poisson variable.e−2⋅233!=43e−2≈0.1804e^{-2} \cdot \frac{2^3}{3!} = \frac{4}{3} e^{-2} \approx 0.1804
4Report it as an approximation. Write ≈\approx. If software is available, compare with the exact binomial value.exact Bin⁡(500,0.004)\operatorname{Bin}(500, 0.004): 0.18080.1808

Worked examples

Example 1 · no successes

k = 0
n=1000n = 1000, p=0.002p = 0.002, P(X=0)P(X = 0)
λ=2\lambda = 2
P(X=0)≈e−2≈0.1353P(X = 0) \approx e^{-2} \approx 0.1353
Poisson
exact: (0.998)1000≈0.1351(0.998)^{1000} \approx 0.1351
close agreement

Example 2 · at most k successes

short sum
n=200n = 200, p=0.01p = 0.01, P(X≤2)P(X \le 2)
λ=2\lambda = 2
e−2(1+2+222)=5 e−2e^{-2}\left(1 + 2 + \frac{2^2}{2}\right) = 5\, e^{-2}
three Poisson terms
P(X≤2)≈0.6767P(X \le 2) \approx 0.6767
exact binomial: 0.67670.6767

Example 3 · at least k successes

complement
n=300n = 300, p=0.01p = 0.01, P(X≥2)P(X \ge 2)
λ=3\lambda = 3
1−e−3(1+3)=1−4 e−31 - e^{-3}(1 + 3) = 1 - 4\, e^{-3}
remove X=0X = 0 and X=1X = 1
P(X≥2)≈0.8009P(X \ge 2) \approx 0.8009
exact binomial: 0.80240.8024
Compare both values in the discrete distributions calculator.

See it happen

Keep the mean npnp fixed and let nn grow: each trial gets rarer, the chance of two successes in one trial vanishes, and the binomial bars settle onto the Poisson bars with λ=np\lambda = np. That limit is the law of rare events. Set λ\lambda in the Poisson distribution visualizer and compare it with a large binomial in the binomial distribution visualizer.

Common mistakes

Using it when p is not small

✗ n=50n = 50, p=0.3p = 0.3: Poisson⁡(15)\operatorname{Poisson}(15)
✓ pp is too large; the binomial variance np(1−p)=10.5np(1-p) = 10.5 is far below 1515. Use the exact binomial or the normal approximation

Wrong λ

✗ λ=n/p\lambda = n/p or λ=p\lambda = p
✓ λ=np\lambda = np, the expected number of successes

Writing an equals sign

✗ P(X=3)=0.1804P(X = 3) = 0.1804 for Bin⁡(500,0.004)\operatorname{Bin}(500, 0.004)
✓ P(X=3)≈0.1804P(X = 3) \approx 0.1804; the exact value is 0.18080.1808