Binomial Distribution Calculator

Need Help with Statistics?

What is the probability of getting at least 8 heads in 10 coin tosses? The answer is 0.0547, and the binomial distribution calculator below gives it in one click, with steps and a graph.

I have spent 12 years helping students and researchers with dissertation statistics. This guide covers the tool, the hand method, and what to do with the number afterwards, which most calculator pages skip.

Use the Binomial Distribution Calculator to Find Binomial Probability Online

Quick answer: For 10 tosses of a fair coin (p = 0.5), P(X ≥ 8) = 0.0547, or about 5.5%. Choose At least, enter n = 10, p = 0.5 and x = 8.

To find binomial probability online, enter n, p and x, pick the type of probability, and press Calculate. Let us see what each input means.

What do n, p and x mean?

• n is the number of trials. It must be fixed before you start, for example 10 tosses or 20 items.
• p is the probability of success on one trial, written as a decimal. Enter 0.25, not 25.
• x is the number of successes the question asks about.

How do I pick exact, at most, at least or between?

For a binomial probability calculator for exact outcomes, choose Exact. Use it when the question says “exactly”, “precisely” or “equal to”.

For everything else, choose a cumulative option. At most covers “x or fewer” and “no more than”. At least covers “x or more”. Between covers a range such as “3 to 6 inclusive”.

This is where a binomial distribution calculator for at least / at most probabilities saves the most time. By hand you must add many terms, and one missed term spoils the answer.

How do I read the binomial distribution calculator with graph and steps?

The binomial distribution calculator with graph draws one bar for each possible number of successes. The shaded bars are the outcomes included in your answer, and their total height is the probability.

The binomial distribution calculator with steps shows the formula with your numbers substituted. Use it to check your own working, not to copy blindly.

Try small values to see the true shape. The plots in the NIST/SEMATECH e-Handbook use n = 100, so they do not show how lopsided small samples look. Enter n = 10 and p = 0.1 here and you will see the skew.

Why use Statssy’s binomial distribution calculator?

• It is free to use.
• It shows the working and the graph, so you can learn the method and not only copy an answer.
• Test it yourself: enter the worked examples in this article and check that the results match.
• For a very large n, check the calculator’s output against R or Excel before relying on it for a live dataset.

Key takeaways

• The binomial distribution counts successes in a fixed number of independent yes/no trials with constant p.
• Check the 4 conditions before you calculate.
• For “at least k”, use 1 − P(X ≤ k − 1).
• The mean is np and the variance is np(1 − p).
• With real survey data, a probability is not enough. You need a binomial test.

What Is a Binomial Distribution?

The binomial distribution gives the probability of exactly k successes in n independent trials, where each trial has only two outcomes and the same probability of success, p. It is written X ~ B(n, p).

Each trial is called a Bernoulli trial, after the Swiss mathematician Jacob Bernoulli. Take 10 coin tosses: each toss is a trial, heads is a success, and p = 0.5. The distribution gives the chance of 0 heads, 1 head, 2 heads and so on, up to 10.

What are the 4 conditions of a binomial distribution?

1. The number of trials n is fixed in advance.
2. Each trial has two outcomes, success or failure.
3. The trials are independent, so one result does not change the next.
4. The probability of success p is the same on every trial.

The NIST/SEMATECH e-Handbook, written for engineers and scientists, states two of these clearly: two mutually exclusive outcomes, and p fixed for all trials.

Omni Calculator’s binomial page is less careful. It explains independence as trials being “mutually exclusive”. That is the wrong term. Mutually exclusive events cannot happen together, while independent trials simply do not influence each other.

To be fair, the very next sentence on that page explains independence correctly. The page also points readers to hypergeometric, negative binomial and normal-approximation tools. My real complaint is in the cumulative section, which I cover below.

When does the binomial distribution not apply?

The usual trap is sampling without replacement. Draw cards from a deck and the deck changes after every draw, so p changes.

Here is how much it matters. Take a lot of 50 pieces with 4 defective (8%), and inspect 20. The chance of at most 1 defective is 0.4716 with the correct hypergeometric model, but 0.5169 if you wrongly use binomial with p = 0.08.

Now take a lot of 5,000 with 100 defective (2%) and inspect the same 20. The two models give 0.9404 and 0.9401, which is practically the same. This is the 10% condition taught in AP Statistics: if the sample is at most 10% of the population, binomial is a safe approximation.

For small populations, use the hypergeometric distribution calculator.

When Should You Use the Binomial Distribution?

Use the binomial distribution when you count how many times something happens in a fixed number of independent yes/no trials with the same probability each time. Typical cases are coin tosses, defective items in a sample, correct guesses on a quiz and yes/no survey answers.

Situation	Use
Fixed number of yes/no trials, independent, same p	Binomial
Draws without replacement from a small population	Hypergeometric
Counting events in a time or space interval, no fixed n	Poisson
Counting trials until the r-th success	Negative binomial (geometric if r = 1)
Large n and p not near 0 or 1	Normal approximation

How to Calculate Binomial Probability by Hand

Knowing how to calculate binomial probability by hand helps in exams and viva (thesis defence) questions. It also helps you catch a wrong calculator input.

What is the binomial probability formula?

P(X = k) = C(n, k) × p^k × (1 − p)^(n − k)

• C(n, k) is the binomial coefficient, read “n choose k”. It counts the ways to choose which k trials are the successes.
• p^k is the probability of k successes.
• (1 − p)^(n − k) is the probability that the remaining trials all fail.

How do you calculate it step by step?

1. Check the four conditions.
2. Write down n, p and k.
3. Work out C(n, k) = n! / (k! × (n − k)!).
4. Work out p^k and (1 − p)^(n − k).
5. Multiply the three parts.

Worked example: exactly 3 heads in 5 tosses. Here n = 5, p = 0.5 and k = 3. C(5, 3) = 10, p^3 = 0.125 and (1 − p)^2 = 0.25.

So P(X = 3) = 10 × 0.125 × 0.25 = 0.3125. Check it in the calculator with the Exact option.

What are the mean, variance and standard deviation?

• Mean (expected value): np
• Variance: np(1 − p)
• Standard deviation: the square root of np(1 − p)

For 10 tosses of a fair coin, the mean is 5 and the variance is 2.5. The standard deviation is about 1.58. In plain words, expect about 5 heads, give or take 1 or 2.

How to Calculate Cumulative Binomial Probability

To calculate cumulative binomial probability, add the exact probabilities up to your value. For “at least” questions, subtract from 1 instead of adding many terms.

What is the difference between PMF and CDF?

The PMF (probability mass function) gives P(X = k), one exact outcome. The CDF (cumulative distribution function) gives P(X ≤ k), everything up to and including k.

Here is n = 5 and p = 0.5:

k	P(X = k), the PMF	P(X ≤ k), the CDF
0	0.0313	0.0313
1	0.1563	0.1875
2	0.3125	0.5000
3	0.3125	0.8125
4	0.1563	0.9688
5	0.0313	1.0000

Binomial distribution calculator for at least and at most probabilities: the cheat table

Wording in the question	Symbol	How to get it
exactly k	P(X = k)	Exact formula
at most k, k or fewer, no more than k	P(X ≤ k)	Add P(0) up to P(k)
fewer than k, less than k	P(X < k)	P(X ≤ k − 1)
at least k, k or more, no fewer than k	P(X ≥ k)	1 − P(X ≤ k − 1)
more than k, greater than k	P(X > k)	1 − P(X ≤ k)
between a and b, inclusive	P(a ≤ X ≤ b)	P(X ≤ b) − P(X ≤ a − 1)

East Central College’s Learning Center publishes a one-page table that maps key phrases to TI-84 commands. The mapping is correct, and as a quick reference it does its job.

My concern is that phrases are a finite list and exam wording is not. “A minimum of 3” or “up to 3” are not on the sheet, even though the sheet does cover “no less than”. The sheet also lists “simple random sample” as a condition but never mentions the 10% check.

My advice: learn the symbol, not the phrase. Ask which values of X the question allows, then write them down.

Why do “at least” answers go wrong?

“At least 5” and “more than 5” are different. Take n = 10 and p = 0.5:

• At least 5 is 1 − P(X ≤ 4) = 0.6230.
• More than 5 is 1 − P(X ≤ 5) = 0.3770.

Subtracting the wrong cumulative value is the classic slip. For “at least k”, always subtract the cumulative value for k − 1.

Omni’s cumulative section only adds terms one by one. It never shows this complement rule or warns about the k − 1 step, which is exactly where students slip.

Worked example: at least 8 heads in 10 tosses

Here n = 10, p = 0.5 and the question is P(X ≥ 8). Adding the three terms gives:

• P(8) = 45 / 1024 = 0.0439
• P(9) = 10 / 1024 = 0.0098
• P(10) = 1 / 1024 = 0.0010

The total is 56 / 1024 = 0.0547. The complement way gives the same: 1 − P(X ≤ 7) = 1 − 0.9453. A fair coin gives 8 or more heads in about 1 experiment out of 18.

Binomial Probability Calculator for Real Problems

Binomial distribution calculator for coin toss problems

A binomial distribution calculator for coin toss problems is the classic use. A fair coin has p = 0.5, and n is the number of tosses. The at-least-8 example above is exactly this.

Coin problems are good for learning, but they make people think binomial is only a classroom topic. The next examples show otherwise.

Binomial distribution calculator for quality control

A binomial distribution calculator for quality control answers a question every factory asks: will this batch pass inspection? The practice is called acceptance sampling.

Illustrative scenario: A workshop packs 20 items in each carton. The buyer rejects any carton with 2 or more defective items. What is the chance a carton is accepted?

Take a defect rate of 2%. Here n = 20 and p = 0.02, so P(X ≤ 1) = 0.6676 + 0.2725 = 0.9401. The chance of rejection is about 6%.

Now vary the defect rate. Plotting acceptance probability against defect rate gives the operating characteristic (OC) curve.

True defect rate	Chance carton is accepted
1%	0.983
2%	0.940
5%	0.736
10%	0.392
15%	0.176

This table is what a buyer really wants to see. It shows how quickly a plan starts rejecting bad lots as quality falls. The NIST handbook is a good source for engineering problems like this, though I find its style dense for first-year students.

Binomial probability for genetics problems

Binomial probability for genetics problems is usually about Mendel. If both parents are Aa, each child has a 0.25 chance of being aa.

Illustrative scenario: A couple with genotype Aa × Aa has 4 children. What is the chance that exactly 1 child is aa?

With n = 4, p = 0.25 and k = 1, the answer is 4 × 0.25 × 0.75³ = 0.4219. The chance of at least one aa child is 1 − 0.75⁴ = 0.6836.

Textbooks give you the parents’ genotypes. Real problems often do not, so you must first estimate the chance that a parent is a carrier, and binomial comes after that. For more than two outcome categories, such as a 9:3:3:1 ratio, use the chi-square goodness of fit calculator.

Binomial probability calculator for survey responses

A binomial probability calculator for survey responses fits any yes/no question, like “Are you satisfied?”.

Illustrative scenario: A company picks 10 customers at random from a very large list. Past data says 70% are satisfied. What is the chance that at least 8 of the 10 say satisfied?

With n = 10 and p = 0.7, P(X ≥ 8) = 0.2335 + 0.1211 + 0.0282 = 0.3828. Even when the true rate is 70%, you would see 8 or more satisfied customers only about 38% of the time.

Real surveys have complications. Respondents from the same household or class are not independent, and if only unhappy people reply, p is not constant. For bigger surveys, the margin of error calculator gives the precision of your estimate.

Binomial Probability Calculator for Students: Check Your Answers in Excel, R, Python and TI-84

A binomial probability calculator for students should let you check your own working. Here is one problem in four tools: n = 20, p = 0.02, and P(X ≤ 1).

Tool	Command	Result
Excel	=BINOM.DIST(1,20,0.02,TRUE)	0.9401
R	pbinom(1, size = 20, prob = 0.02)	0.9401
Python	from scipy.stats import binom then binom.cdf(1, 20, 0.02)	0.9401
TI-84	binomcdf(20,0.02,1)	0.9401

For an exact probability, use BINOM.DIST(k,n,p,FALSE) in Excel, dbinom in R and binom.pmf in Python. For “at least 8 in 10”, type =1-BINOM.DIST(7,10,0.5,TRUE) in Excel to get 0.0547.

Stuck on which test your data needs? Send your question to Statssy on WhatsApp.

Pearson’s calculator page offers sensible quick picks, including survey and defect examples, and a normal-approximation note. But its worked examples are a true/false quiz and free throws, and the survey pick stops at a probability. It never tells a student what to do when they hold real survey counts, which is why the next section exists.

For UK A-level and US AP Statistics

For UK A-level, you write X ~ B(n, p). Save My Exams’ Edexcel notes point out that when p is bigger than 0.5 you can count failures instead, using B(n, 1 − p). This helps when you work from printed cumulative tables. Many scientific calculators have binomial functions, but do check your exam board’s rules.

For US AP Statistics, graphing calculators with binompdf and binomcdf are standard, but you must justify the model. The usual check is the 10% condition, and Khan Academy has a short video on the 10% rule. A bare number with no conditions checked may lose marks.

From Calculator to Dissertation: The Binomial Test

The pages I reviewed stop at the probability. Your supervisor will not ask for P(X = 26). They will ask if your result differs from chance, and the binomial test answers that.

When does survey data need a binomial test?

Use it when you have one yes/no variable, a smallish sample and a benchmark proportion p₀. The hypotheses are H₀: p = p₀ against H₁: p ≠ p₀ (two-tailed). For large samples, the one-proportion z-test calculator gives a close answer.

Illustrative scenario: An MBA student surveys 40 users. 26 prefer the new app design, which is 65%. Her benchmark is 50%, meaning no preference.

The exact two-sided p-value is 0.081. That is above 0.05, so the result is not statistically significant, even though 65% sounds strong. With 40 people, this can easily happen by chance.

A note on one-tailed tests. If she had claimed before the survey that the new design would be preferred, the one-sided p-value would be 0.040. Choosing the direction after seeing the data is not honest, so decide it in advance and defend it.

To run it in SPSS, go to Analyze > Nonparametric Tests > Legacy Dialogs > Binomial. In R, use binom.test(26, 40, p = 0.5), and in Python scipy.stats.binomtest(26, 40, 0.5). For a benchmark other than 0.5, let the software choose the two-sided method, and do not just double the tail.

Illustrative scenario: A nursing dissertation audited fall-risk screening compliance before and after a ward safety huddle was introduced. The trust’s historical baseline was 70% compliance. Out of 80 patient records audited after the intervention, 68 showed a completed screening, which is 85%.

An exact binomial test against p₀ = 0.70 gives a two-sided p-value of 0.003, so the improvement is unlikely to be due to chance alone. The 95% Clopper-Pearson interval for the new compliance rate is 75.3% to 92.0%, and the dissertation would report both the p-value and this interval, not one alone.

Compare this with the app-design example above. Both surveys found a higher proportion than the benchmark, 65% and 85%. Only the second one is statistically significant, because the sample was twice as large and the gap from the benchmark was wider. This is exactly why the power section below matters.

How many respondents do you need?

This also shows why sample size matters. If the true preference is 65% against a 50% benchmark, an exact two-sided test at the 5% level has only about 44% power with n = 40. It needs about 90 respondents to reach roughly 81% power. Plan this before collecting data with the sample size and power calculator.

How do you report a binomial test result?

A sample sentence in APA style:

An exact binomial test showed that the proportion of respondents preferring the new design (0.65, 95% CI [0.48, 0.79]) did not differ significantly from 0.50, p = .081 (two-tailed).

For the exact p-value itself, R’s binom.test above is the most direct route. If you want to cross-check it, see whether the p-value calculator for all distributions covers the exact binomial test. To decide which test fits your data in the first place, use our which statistical test to use guide.

Always report a confidence interval, and name its method. For 26 of 40, the Clopper-Pearson (exact) interval is 0.48 to 0.79 and the Wilson interval is 0.50 to 0.78. Use the confidence interval for one proportion calculator for a quick interval, and check which method it uses (Wald, Wilson or exact) before you quote it, since this article recommends against plain Wald for small samples.

This choice deserves a critique. Brown, Cai and DasGupta (2001) showed that the standard Wald interval has erratic coverage, and they argue that common textbook rules about when it is safe cannot be trusted. For small n they recommend the Wilson or Jeffreys interval, and Agresti-Coull for larger n.

I agree with the message. My criticism is only the delivery: the paper is written for statisticians, so many students never meet it, and in my view, many students meet the Wald interval first in their coursework.

Clopper and Pearson (1934) gave the “exact” interval, which guarantees coverage but tends to be wide. Agresti and Coull (1998) argued that approximate intervals can be better than exact ones. In a dissertation, I would report Wilson or exact, and say which.

Common Mistakes, and When Binomial Is the Wrong Tool

What are the six most common mistakes?

1. Entering p as a percentage. Use 0.25, not 25.
2. Rounding p too early. Use 1/3 or 0.3333, not 0.33, and round only at the end.
3. Trials that are not fixed. “Tosses until the first head” is geometric, not binomial.
4. Ignoring clustering. Students from the same class, or patients from the same clinic, are not independent.
5. Treating a probability as a test result. P(X = 26) is not a p-value. You need the tail probability and the hypotheses.
6. Using an estimated p as if it were known. The sample proportion p̂ = x/n carries uncertainty.

Normal, Poisson or hypergeometric: which approximation should I use?

• Normal approximation: use when np ≥ 10 and n(1 − p) ≥ 10, with a continuity correction of 0.5. For n = 100 and p = 0.5, P(X ≥ 60) is 0.0284 exactly, 0.0287 with the correction, and 0.0228 without it.
• Poisson approximation: use when n ≥ 100 and np ≤ 10, with λ = np. For n = 200 and p = 0.01, P(X ≤ 1) is 0.4046 exactly and 0.4060 by Poisson. Check it with the Poisson distribution calculator.
• Hypergeometric: use for small populations, as shown earlier.

Statistics Kingdom’s binomial page is feature-rich, with inverse calculation, a draggable chart, and R and Excel code. Its text says the normal approximation works when n is large and p is not too close to 0 or 1, but it gives no numbers.

A beginner who switches on the overlay at n = 10 and p = 0.1 will trust a curve that fits badly. There, the normal curve gives 0.943 for P(X ≤ 2) against the exact 0.930. Please check np first.

When to Get Expert Help

You can use this binomial distribution calculator to check answers and learn the method. But if your dissertation needs the right test, correct assumptions and a clean write-up, a second pair of eyes saves weeks.

That is the work I do at Statssy. Send me your research question and a description of your data, and I will tell you whether a binomial test, a proportion test or something else fits. Talk to Statssy on WhatsApp about your dissertation statistics.

Frequently Asked Questions

About the author

Siddharth is a researcher with an MBA (Finance) and an M.Tech, and 12 years of experience in dissertation support and applied statistics. He works with R, Python, SPSS and Stata and runs Statssy. Connect with him on LinkedIn.