BINOM.INV in Google Sheets: Find a Binomial Count Threshold

=BINOM.INV(num_trials, prob_success, target_prob)

• num_trials: the fixed number of independent trials.
• prob_success: probability of a success in one trial.
• target_prob: the cumulative probability threshold to reach.

All three arguments are supplied; no optional defaults are documented. Google lists BINOM.INV as the name pointing to CRITBINOM. The contract is the smallest count for which cumulative probability is greater than or equal to the threshold.

The output is a count, not a probability and not an estimated number of future successes. Keep trials whole and nonnegative. For the examples, success and target probabilities are strictly between 0 and 1; the direct help does not give a comprehensive endpoint, invalid-input or coercion specification.

Enter the following parameters in A1:B4:

In D2 enter:

=BINOM.INV(B2,B3,B4)

Expected result: 3. You can understand the answer without treating the function as a black box:

Each cumulative value adds the exact count probabilities from zero upward. The first row to meet 0.8 is the count 3 row. There is no attainable cumulative value exactly equal to 0.8, so the function passes the target rather than interpolating a fractional count.

Put counts 0, 1, 2, 3, 4 in F2:F6. Enter this in G2 and fill down to G6:

=BINOMDIST(F2,$B$2,$B$3,TRUE)

The expected G2:G6 values are the probability ladder above. Google’s BINOMDIST reference defines TRUE as the inclusive cumulative mode. The dollar signs keep the shared trial count and probability fixed while the tested count changes.

Check the selected result and its predecessor in D3 and D4:

=BINOMDIST(D2,$B$2,$B$3,TRUE)
=BINOMDIST(D2-1,$B$2,$B$3,TRUE)

Expected results: 0.9375 and 0.6875. The selected count meets the target; the preceding count does not. This predecessor check assumes the selected count is positive. Do not mechanically call a count distribution with −1 when the quantile is zero.

Change B4 to 0.6875. The expected BINOM.INV answer is 2, because equality already satisfies the threshold.

Change B4 to 0.69. The expected answer becomes 3. This abrupt change is a feature of a discrete distribution. If the display rounds both thresholds to “69%,” increase the displayed decimal precision before diagnosing a problem.

Do not round the target probability as a preprocessing step unless that is genuinely part of the model. A small change near a cumulative jump can alter the integer answer.

Suppose eight independent sites may each request one replacement part, each with an assumed 25% request probability. To choose the smallest stock count covering at least 90% of this model’s demand outcomes:

=BINOM.INV(8,0.25,0.9)

Expected result: 4 parts. Independent calculations give:

• P(demand ≤ 3) = 0.8861846923828125, below 90%.
• P(demand ≤ 4) = 0.9727020263671875, meeting 90%.

The mean demand is only 8×0.25 = 2. The mean and a high-service-level capacity threshold answer different questions. Here the discrete jump makes actual coverage about 97.27%, even though the requested minimum was 90%.

This example assumes one possible part per site, independent requests and a common probability. Multiple-part requests, shared failures or changing risk require a different demand model. BINOM.INV does not evaluate costs, estimate probabilities or guarantee the service level will occur in a particular period.

• Probability and target swapped: Both are on the 0–1 scale, so a formula can look plausible while answering a different question. Label the inputs.
• Expecting the target as output: The result is the threshold count. Use a cumulative probability function to inspect its achieved coverage.
• Comparing with “greater than” instead of “greater than or equal”: Equality belongs to the selected count, as the 0.6875 example demonstrates.
• Imported percentages: 90 is not 90%. Keep numeric probabilities separate from display formatting.
• Fractional trial count: Choose an explicit whole-count policy; the reviewed help does not establish which coercion or error to expect.

For an error, first check the three numeric inputs and local formula separators, then reduce to the four-trial example. The cited Google pages do not enumerate a complete native error-code matrix, so this guide does not borrow error outcomes from Excel or another spreadsheet engine.

Google’s current documentation lists the BINOM.INV name and its CRITBINOM reference. No original introduction date, launch version or last functioning numbered release was established. This is the ordinary rolling Sheets worksheet service.

Evidence status: NOT_TESTED_IN_GOOGLE_SHEETS. Sources were inspected on October 17, 2026. The cumulative ladders and threshold selections were independently calculated using exact binomial arithmetic. No native execution, account-specific availability check or screenshot is claimed. When migrating a workbook, compare the count, selected cumulative value and predecessor rather than trusting a matching function name alone.

Related published guides: COMBIN, PROB and COUNT.