Written and maintained by CASRAI Editorial Board
Published
On this page
The Poisson cumulative distribution function (CDF) calculates the probability of observing at most k independent event occurrences within a fixed interval, given a constant mean rate λ. Rather than evaluating an exact count, the cumulative distribution sums the individual Poisson probabilities from zero up to k. Researchers use the cumulative distribution to assess operational thresholds, evaluate rare-event frequencies, and determine tail risks. For related foundations, see our types of distributions in statistics reference or our foundational Poisson distribution guide.
Ask CASRAI · free to tryQuestion about this guide? Ask CASRAI answers with linked sources. 2 free questions a day, no account.
Definition in one paragraph: Poisson cumulative distribution
The cumulative distribution of a Poisson random variable measures the aggregate probability that an observed count X does not exceed an integer threshold k. While the probability mass function (PMF) calculates the likelihood of observing exactly k occurrences, the cumulative distribution function (CDF) calculates the likelihood of observing k or fewer occurrences. Because Poisson models govern discrete count data, the CDF is a step function: it remains constant between integers and jumps at each non-negative integer k by that count’s probability mass, starting at zero for negative values and approaching one as k increases.
Formula with every symbol explained: Poisson CDF
The Poisson cumulative distribution function, denoted as F(k; λ) or P(X ≤ k), equals the sum of the discrete probability mass terms from zero to k:
F(k; λ) = P(X ≤ k) = ∑i=0k (e-λ × λi) / i!
Each mathematical symbol in this cumulative expression has a defined operational meaning:
- X: The discrete random variable representing the observed count of event occurrences within the interval.
- k: The upper threshold count of interest (a non-negative integer: k = 0, 1, 2, …). Non-integers supplied to analytical software are truncated to ⌊k⌋.
- λ (lambda): The rate parameter representing the expected mean count of occurrences per interval (λ > 0). In a Poisson distribution, λ equals both the theoretical mean (μ) and theoretical variance (σ2). The standard deviation is √λ.
- e: Euler’s constant (≈ 2.71828).
- i!: The factorial of integer i (with 0! = 1).
- ∑i=0k: The summation operator, directing the addition of terms from i = 0 through i = k.
In analytical software, the Poisson CDF is often evaluated via the regularized incomplete gamma function, Q(⌊k + 1⌋, λ) = Γ(⌊k + 1⌋, λ) / ⌊k⌋!, avoiding factorial overflow during numerical computation.
Worked example with real numbers and arithmetic shown
Consider a computing facility that averages λ = 3.0 unplanned node failures per week. The administrator wants to find the probability that at most two node failures occur during a given week: P(X ≤ 2).
Using the cumulative summation formula with λ = 3.0 and k = 2, first evaluate the exponential term e-3.0:
e-3.0 ≈ 0.049787
Next, evaluate the individual probability mass terms for each integer count from i = 0 to i = 2:
- For i = 0: P(X = 0) = (e-3.0 × 3.00) / 0! = (0.049787 × 1) / 1 = 0.049787
- For i = 1: P(X = 1) = (e-3.0 × 3.01) / 1! = (0.049787 × 3.0) / 1 = 0.149361
- For i = 2: P(X = 2) = (e-3.0 × 3.02) / 2! = (0.049787 × 9.0) / 2 = 0.224042
Summing these terms yields the cumulative probability:
P(X ≤ 2) = 0.049787 + 0.149361 + 0.224042 = 0.423190
The cumulative probability of observing at most two failures in a week is 0.4232 (42.32%). The complementary probability of observing more than two failures is P(X > 2) = 1 – P(X ≤ 2) = 1 – 0.423190 = 0.576810 (57.68%). For automated lookups, see our Poisson distribution solver.
Interpretation: what the result means, with a table of reading guidance
Interpreting a cumulative Poisson probability requires mapping research questions to correct inequalities. In discrete distributions, strict and non-strict inequalities produce different outcomes.
| Research Question | Inequality Expression | Equivalent Poisson CDF Calculation | Example (λ = 3.0, k = 2) |
|---|---|---|---|
| At most k / No more than k | P(X ≤ k) | F(k; λ) | P(X ≤ 2) = 0.4232 |
| Fewer than k / Less than k | P(X < k) | F(k – 1; λ) | P(X ≤ 1) = 0.1991 |
| At least k / No fewer than k | P(X ≥ k) | 1 – F(k – 1; λ) | 1 – P(X ≤ 1) = 0.8009 |
| More than k / Strictly exceeding k | P(X > k) | 1 – F(k; λ) | 1 – P(X ≤ 2) = 0.5768 |
| Between a and b inclusive | P(a ≤ X ≤ b) | F(b; λ) – F(a – 1; λ) | P(1 ≤ X ≤ 2) = 0.4232 – 0.0498 = 0.3734 |
A cumulative probability below 0.05 indicates an unusually low count under rate λ, whereas an upper-tail probability 1 – F(k; λ) below 0.05 indicates an unusually large event cluster.
Poisson cumulative distribution table
The table below provides reference cumulative probabilities F(k; λ) = P(X ≤ k) across standard rate parameters λ ∈ {0.5, 1.0, 2.0, 3.0, 5.0} for counts k = 0 through 5, calculated from NIST formulas.
| Count (k) | λ = 0.5 | λ = 1.0 | λ = 2.0 | λ = 3.0 | λ = 5.0 |
|---|---|---|---|---|---|
| 0 | 0.6065 | 0.3679 | 0.1353 | 0.0498 | 0.0067 |
| 1 | 0.9098 | 0.7358 | 0.4060 | 0.1991 | 0.0404 |
| 2 | 0.9856 | 0.9197 | 0.6767 | 0.4232 | 0.1247 |
| 3 | 0.9982 | 0.9810 | 0.8571 | 0.6472 | 0.2650 |
| 4 | 0.9998 | 0.9963 | 0.9473 | 0.8153 | 0.4405 |
| 5 | 1.0000 | 0.9994 | 0.9834 | 0.9161 | 0.6160 |
When λ = 0.5, the cumulative probability reaches 0.9856 by k = 2. As λ rises to 5.0, probability mass disperses rightward, with P(X ≤ 2) dropping to 0.1247.
Assumptions and common mistakes
Applying the Poisson cumulative distribution requires four foundational assumptions:
- Discrete non-negative integer counts: Events must be indivisible discrete instances (0, 1, 2, …).
- Event independence: The occurrence of one event must not alter the likelihood of subsequent events.
- Constant rate parameter (λ): The occurrence rate must remain stable across the observation window.
- Equidispersion (μ = σ2): Theoretical variance must equal the mean. When observed variance exceeds the mean (overdispersion), the Poisson CDF underestimates tail risks; in such cases, consult our comparison of quasi-Poisson vs negative binomial models.
Three common procedural mistakes arise in practice:
- Off-by-one tail indexing: Formulating “at least k events” as 1 – F(k) instead of 1 – F(k – 1), which mistakenly excludes count k.
- Rate and window mismatch: Applying a rate defined over one timeframe to a different observation window without scaling λ proportionally.
- Treating discrete sums as continuous integrals: Omitting continuity corrections when using normal approximations for Poisson cumulative counts.
How to report it (APA example)
In APA 7th style, report the assumed arrival rate λ, the observation window, the count threshold, and the exact probability. Parameters (λ, M, SD, p) must be italicized, and probabilities bounded by one are reported to three decimal places without leading zeros.
An APA-formatted reporting example:
“Unscheduled equipment service interruptions were modeled using a Poisson distribution with an empirical mean arrival rate of λ = 3.00 interruptions per 40-hour operational shift (SD = 1.73). The cumulative probability of experiencing at most two interruptions during a single shift was P(X ≤ 2) = .423. Conversely, the probability of exceeding the operational buffer of two interruptions was P(X ≥ 3) = .577. A chi-square goodness-of-fit test confirmed that observed interruption frequencies did not deviate significantly from theoretical cumulative Poisson expectations, χ2(4, N = 200) = 3.42, p = .490.”
For Bayesian parameter estimation and prior specifications on λ, see our guide to Bayesian regression.
How to calculate it in Excel, R or Python
Standard analytical packages provide verified functions to evaluate the Poisson cumulative distribution directly.
Microsoft Excel
In Excel, use POISSON.DIST with the cumulative argument set to TRUE:
=POISSON.DIST(2, 3.0, TRUE)
This returns 0.423190. Setting the argument to FALSE returns the single-count PMF. For upper tail P(X ≥ 3), enter =1 - POISSON.DIST(2, 3.0, TRUE).
R (stats package)
In R, function ppois() evaluates the cumulative distribution. Use lower.tail = FALSE for upper tail probabilities to avoid floating-point cancellation:
# Lower tail: P(X <= 2) with lambda = 3.0
ppois(q = 2, lambda = 3.0, lower.tail = TRUE)
# Output: [1] 0.4231901
# Upper tail: P(X >= 3) = P(X > 2) with lambda = 3.0
ppois(q = 2, lambda = 3.0, lower.tail = FALSE)
# Output: [1] 0.5768099
Python
In Python, calculate cumulative sums using the standard library math module or scipy.stats:
import math
def poisson_cdf(k, lam):
return sum(math.exp(-lam) * (lam ** i) / math.factorial(i) for i in range(k + 1))
print(f"P(X <= 2): {poisson_cdf(2, 3.0):.6f}")
# Output: P(X <= 2): 0.423190
# SciPy: stats.poisson.cdf(2, 3.0); stats.poisson.sf(2, 3.0) for upper tail
FAQ
What is the difference between a Poisson PMF and a Poisson CDF?
The Poisson probability mass function (PMF) calculates the exact probability of observing a single integer count k: P(X = k). In contrast, the Poisson cumulative distribution function (CDF) calculates the aggregate probability of observing any count less than or equal to k: P(X ≤ k) = ∑ P(X = i) for i = 0 to k.
How do you calculate the probability of at least k events in a Poisson distribution?
To calculate the probability of observing at least k events, apply the complement rule to the CDF evaluated at k – 1: P(X ≥ k) = 1 – P(X ≤ k – 1). For example, with λ = 3.0, the probability of at least 3 events is 1 – P(X ≤ 2) = 1 – 0.4232 = 0.5768.
Can the rate parameter lambda be a decimal in the Poisson cumulative distribution?
Yes. The rate parameter λ represents an average expected occurrence count per interval and can be any positive real number (λ > 0), such as 0.75 or 4.2. In contrast, the event count k must always be a non-negative integer (0, 1, 2, …).
Why do the mean and variance equal lambda in a Poisson cumulative distribution?
In a standard Poisson process, arrival events occur independently at a constant mean rate. This independence property constrains the variance to equal the mean: μ = σ2 = λ. When empirical data exhibits variance significantly greater than the mean, overdispersion is present.
Sources
- National Institute of Standards and Technology. (2026). 1.3.6.6.19. Poisson Distribution. NIST/SEMATECH e-Handbook of Statistical Methods. https://www.itl.nist.gov/div898/handbook/eda/section3/eda366j.htm (Accessed November 8, 2026).
- R Core Team. (2026). The Poisson Distribution (R stats package documentation). Comprehensive R Archive Network / ETH Zurich. https://stat.ethz.ch/R-manual/R-devel/library/stats/html/Poisson.html (Accessed November 8, 2026).
- SciPy Community. (2026). scipy.stats.poisson: SciPy v1.18.0 Reference Guide. SciPy Documentation. https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.poisson.html (Accessed November 8, 2026).
- Microsoft Corporation. (2026). POISSON.DIST function. Microsoft 365 Support. https://support.microsoft.com/en-us/office/poisson-dist-function-8d14848f-7dd0-4527-a0e3-b83748d52163 (Accessed November 8, 2026).







