ERFC.PRECISE Function in Google Sheets: Formula, Examples & Troubleshooting

=ERFC.PRECISE(z)

Google’s function catalog lists this exact spelling and its one-argument signature, then directs readers to the ERFC reference. That reference identifies the argument as the numeric input to the complementary Gauss error function.

There is one required argument and no optional upper limit in the documented ERFC.PRECISE signature. The catalog cross-reference establishes the documented mathematical purpose. It does not establish that the two names use an identical implementation, have a measured speed difference, or return bit-for-bit identical native results.

Mathematically, the complementary error function is erfc(z) = 1 − erf(z). It is also the Gaussian integral from z to positive infinity, multiplied by 2/√π. This is a special mathematical function; its raw output is not automatically a standard-normal probability without a transformation.

Enter these test inputs in A2:A7, with z in A1 and Complementary error function in B1:

1. Put =ERFC.PRECISE(A2) in B2.
2. Fill down to B7.
3. Show enough decimal places to distinguish the small positive values from zero.
4. Compare with these independently calculated mathematical expectations:

The decline as z increases is a property of the defining integral. A value above 1 for a negative input is not inherently a mistake: the complementary error function spans values between 0 and 2, approaching its endpoint limits as z moves toward positive or negative infinity.

As an independent consistency check, the mathematical identity erfc(-z) + erfc(z) = 2 gives about 2 for the rows with z = -1 and z = 1. This identity checks the expected arithmetic; it is not a native test result.

Suppose a synthetic measurement model has mean 100 and standard deviation 5. You want the model’s probability of a value above 110. Set D2 to 100, E2 to 5, and F2 to 110.

A normal-model right-tail formula is:

=0.5*ERFC.PRECISE((F2-D2)/(E2*SQRT(2)))

Here the standardized distance is (110 − 100)/5 = 2. The argument passed to ERFC.PRECISE is therefore 2/√2, not 2. The expected probability is approximately 0.02275013195, or 2.275013195%.

Two common arithmetic mistakes have visibly different results. Forgetting the leading 0.5 doubles the probability; passing the standardized distance without dividing by √2 uses the wrong scale. Keep mean, standard deviation, and threshold in the same measurement unit, and require a positive standard deviation for this model.

The formula describes an assumed normal distribution. It does not establish that real measurements follow that distribution or justify a consequential decision from the probability alone. With a threshold at the mean, the same derivation gives 0.5 because erfc(0) = 1.

• A negative input gives more than 1: Check whether you intended the raw complementary error function or a transformed normal probability.
• A small tail displays as zero: Increase the displayed precision or use scientific notation before concluding the stored value is exactly zero.
• A formula includes two bounds: ERFC.PRECISE has one documented argument. A difference between two complementary-function values is a separate mathematical construction.
• The input is a label or imported text: Check the source cell. Google’s linked reference does not specify all text-coercion cases or exact error tokens, so this guide does not promise automatic conversion.
• You are comparing precision: The suffix alone is not a numerical benchmark. No comparative accuracy or implementation claim is made here.

Computing a tiny tail directly with a complementary function also avoids writing an explicit subtraction of nearly equal displayed values such as 1 − 0.999999…. That is a mathematical motivation for the formulation, not a measured guarantee about Sheets’ internal algorithm.

The exact ERFC.PRECISE name is in Google’s current catalog as reviewed. Its introduction date, any retirement version, and parity with other spreadsheet implementations were not established. Keep the documented spelling when moving a formula between files, and verify the destination product separately.

See SQRT for the normal-tail scaling, EXP for exponential models, and ROUND for explicit output rounding.

First-party product references reviewed October 17, 2026: Google Sheets function catalog and ERFC. Mathematical examples were checked independently using the defining function and normal-tail identity.