Poisson Distribution Calculator — Exact & Cumulative Probability

Poisson Distribution Calculator

Find the probability of a given number of events in a fixed interval—exactly, at most, at least, or within a range—and the count you will not exceed with a chosen level of confidence.

λ must match your interval

λ is the average number of events in the interval you ask about. If you know a rate for a different period—say 3 per hour but you want an 8-hour shift—switch to Rate × Interval and the calculator scales it for you.

What do you want to calculate?

Average number of events

How to Use This Calculator

  1. Pick a mode: Single Value gives P(X = x) together with all four cumulative probabilities; Range gives the chance the count falls between two values; Count for Target finds the smallest count that covers a chosen share of intervals.
  2. Enter the average λ for your interval, or switch to Rate × Interval and enter a rate per unit and the number of units.
  3. Fill in x, the range a to b, or the target probability.
  4. Press Calculate. The bar chart highlights the counts included, and the step list shows e−λ, λx, and x! as used in the formula.

What Is the Poisson Distribution?

A Poisson distribution calculator calculates the probability that an event occurs a specific number of times within a fixed interval. The Poisson distribution uses the average event rate and assumes events occur independently. Enter the expected event rate and number of occurrences to calculate the probability of that outcome.

The distribution covers counts of 0, 1, 2, 3, … events in a fixed interval of time or space, with no upper limit. It is named after the French mathematician Siméon Denis Poisson, who described it in 1837.

Typical examples are calls reaching a help desk per hour, typos per page, defects per square meter of fabric, accidents at an intersection per year, or radioactive decays per second. In each case you count how many times something happens, with no fixed upper limit, rather than how many out of a fixed number of tries succeed.

Independent

One event does not make another more or less likely.

Constant rate

The average rate stays the same across the interval.

One at a time

Two events cannot happen at exactly the same instant.

Proportional

Doubling the interval doubles the expected count.

Background reference: Encyclopaedia Britannica — Poisson distribution.

Poisson Distribution Formula

Exact probability

P(X = x) = λx · e−λ ÷ x!

x = 0, 1, 2, …

Cumulative probability

P(X ≤ x) = e−λ · Σk=0..x λk ÷ k!

P(X ≥ x) = 1 − P(X ≤ x − 1)

e ≈ 2.71828 is Euler's number and x! is x factorial (x × (x − 1) × … × 1, with 0! = 1). Each probability can also be built from the previous one: P(X = x) = P(X = x − 1) × λ ÷ x. For large counts the calculator works with logarithms and evaluates cumulative sums through the regularized incomplete gamma function, so it stays accurate for λ in the hundreds of thousands.

MeasureValue
Meanλ
Varianceλ
Standard deviation√λ
Mode⌊λ⌋ (both λ − 1 and λ when λ is a whole number)
Skewness1 ÷ √λ
Excess kurtosis1 ÷ λ

Software reference: R documentation — The Poisson Distribution (dpois, ppois, qpois).

Worked Example: Live Chat Requests

A website's support team receives an average of 2.5 chat requests per hour (λ = 2.5). First, e−2.5 = 0.0821.

  • Exactly 4 chats in an hour: 2.5⁴ × 0.0821 ÷ 4! = 39.0625 × 0.0821 ÷ 24 ≈ 0.1336, about 13%.
  • At most 2 chats: 0.0821 × (1 + 2.5 + 3.125) ≈ 0.5438, so a quiet hour is slightly more likely than a busy one.
  • 5 or more chats: 1 − P(X ≤ 4) ≈ 1 − 0.8912 = 0.1088, roughly one hour in nine.
  • Staffing: capacity for 5 chats an hour covers 95.8% of hours; capacity for 4 covers only 89.1%.
  • Full 8-hour shift: λ = 2.5 × 8 = 20, and 28 chats cover 96.6% of shifts.

Chance of At Least One Event

The probability of zero events is e−λ, so the probability of one or more is 1 − e−λ. This quick check answers questions such as "how likely is at least one failure this month?"

Expected events λP(none)P(at least one)
0.190.5%9.5%
0.560.7%39.3%
136.8%63.2%
213.5%86.5%
35.0%95.0%
50.7%99.3%

An event expected once per period (λ = 1) still fails to appear 36.8% of the time. The λ = 3 row also explains the rule of three: if you observe zero events in n units, the 95% upper limit for the true rate is about 3 ÷ n, because a higher rate would make "none at all" less than 5% likely.

Rule-of-three reference: Hanley & Lippman-Hand — If nothing goes wrong, is everything all right? (JAMA, 1983).

Poisson, Binomial, Normal, and Negative Binomial

The Poisson distribution sits between several other models. Knowing the links tells you when it is a good shortcut and when it is the wrong tool.

Approximates the binomial

With many trials n and a small success chance p, binomial counts are close to Poisson with λ = np. A common rule of thumb is n ≥ 20 and p ≤ 0.05. Example: 2 defects among 100 parts at p = 0.02 is 0.2734 exactly and 0.2707 with Poisson.

Approaches the normal

For λ above about 10, the shape is close to a normal curve with mean λ and standard deviation √λ. Add 0.5 to x as a continuity correction. For λ = 20, P(X ≤ 25) is 0.8878 exactly and 0.8906 by the normal approximation.

Linked to the exponential

If counts per hour are Poisson with rate r, the waiting time between events follows an exponential distribution with mean 1 ÷ r. At 2.5 chats per hour, the average gap is 24 minutes.

Replaced by the negative binomial

A Poisson model forces the variance to equal the mean. If your data's variance is clearly larger—common with accidents, insurance claims, or website visits—the negative binomial distribution fits better.

When not to use the Poisson distribution

Avoid it when the rate changes over the interval (rush hours, seasons), when events cluster or trigger each other (aftershocks, contagion, viral posts), when there is a hard maximum (seats on a plane), or when the count is really successes out of a fixed number of trials. In those cases split the interval, use a binomial or negative binomial model, or model the rate directly.

Frequently Asked Questions

What does λ (lambda) mean?

λ is the distribution's only parameter: the average rate of events in the interval you are studying, which is also the expected value of the count. It is both the mean and the variance, so the standard deviation is √λ. It does not have to be a whole number—2.5 calls per hour is a valid λ even though the random variable itself, the number of calls in a given hour, is always a whole number.

How do I convert a rate to a different time period?

Multiply the rate by the length of the new time interval in the same unit. 12 calls per hour becomes 12 × 0.25 = 3 calls per 15 minutes, or 12 × 8 = 96 per 8-hour shift. The same works for a space interval: 0.4 defects per square meter becomes 2 defects per 5 m². Do not scale the probabilities themselves; recalculate them with the new λ as the input.

What is the difference between P(X ≤ x) and P(X < x)?

P(X ≤ x) is the cumulative distribution function and includes x itself; P(X < x) stops at x − 1. Because the Poisson distribution is a discrete distribution with whole-number counts, the difference is exactly P(X = x)—the value of the probability mass function at x—which can be large. "At most 4" means ≤ 4, "fewer than 4" means < 4.

How do I check whether my data follows a Poisson distribution?

Count events in many equal intervals of your dataset and compare the sample mean with the sample variance; for Poisson data they should be similar. Then compare how often you observed 0, 1, 2, … events—the observed frequency of each count—with the probabilities from this calculator. A large excess of zeros or of very high counts in your count data points to another probability model, and a larger sample gives a more reliable estimate of the population rate.

How can I use the calculator for staffing or capacity?

Use Count for Target. Enter the average demand per interval and the share of intervals you want to cover, such as 0.95. The output is the smallest capacity that handles demand that often, and the result also shows the coverage one unit lower.

Why are there sometimes two modes?

When λ is a whole number, P(X = λ − 1) and P(X = λ) are exactly equal. The formula λx · e−λ ÷ x! means each probability equals the previous one multiplied by λ ÷ x, because the factorial gains one factor of x at each step; at x = λ that ratio is 1. For λ = 3, finishing with 2 or 3 events is equally likely.

Can λ be 0?

Only as a trivial case: with λ = 0, the exponential function gives e0 = 1, so zero events is the only possible outcome. The calculator requires λ greater than 0. If you saw no occurrences at all, use the rule of three in the section above to put an upper limit on the rate instead.

Who first applied the Poisson distribution to real data?

The best-known early case is Ladislaus Bortkiewicz's 1898 study of soldiers in the Prussian cavalry killed by horse kicks. It was not a planned experiment, but deaths per corps per year matched Poisson probabilities closely, showing that rare, independent events follow the pattern even when each one seems random.

Other Useful Calculators

Disclaimer: This calculator provides statistical results for educational, informational, and preliminary analytical use. The Poisson model assumes that events occur independently, one at a time, at a constant average rate across the interval; results may be misleading when rates vary, events cluster, or counts have a fixed maximum. Calculations depend entirely on the values entered, and an average rate estimated from limited data carries its own uncertainty. For staffing, safety, insurance, medical, engineering, or other consequential decisions, check the model against observed data and consult a qualified statistician or subject-matter professional.

Last updated: October 4, 2026