RRI Function in Google Sheets: Formula, Examples & Troubleshooting

=RRI(number_of_periods, present_value, future_value)

All arguments are required. number_of_periods is the elapsed number of periods, and the two values are the starting and ending amounts. Google requires the period count and present value to be greater than zero. Although its notes broadly say values must be positive, they explicitly allow a zero future value and give a result of −1, meaning −100%. Google RRI reference.

For positive endpoints, the mathematical solution is (future_value/present_value)^(1/number_of_periods)-1. The rate is per supplied period; the function does not infer years from dates.

Enter this row:

1. Enter =RRI(A2,B2,C2) in D2. Expected result: 10%.
2. Enter =B2*(1+D2)^A2 in E2. Expected reconstructed amount: 13310.
3. To see the path, calculate successive amounts: 11000, 12100, 13310.

The total endpoint increase is 33.1%, but dividing 33.1% by three gives approximately 11.033333% and ignores compounding. RRI solves for the constant compounded rate, not the simple average of a total increase.

Label D2 as annual only if A2 represents years. If the observations are three months apart, the result is 10% per month, a very different statement.

Set A5 to 24, B5 to 10000, and C5 to 12100, representing two years of monthly periods. =RRI(A5,B5,C5) in D5 has expected result approximately 0.7974140429% per month.

In E5 enter =(1+D5)^12-1. Expected effective annual rate: 10%. Multiplying the monthly rate by twelve instead gives approximately 9.5689685147%, a nominal annualization rather than the same compounded annual measure.

This example treats the starting and ending values as a no-flow growth relationship. If an additional deposit caused part of the increase, the endpoint ratio alone cannot separate that deposit from investment growth. A cash-flow-aware model is needed; RRI does not accept a series of deposits or dates.

=RRI(5,10000,20000) has expected result approximately 14.8698354997% per period. It answers what constant rate would double the amount in five periods. It does not show that such a return is achievable or appropriate.

For a decline, =RRI(2,100,81) has expected result −10%. The check is 100*0.9^2 = 81. The endpoints remain positive even though the returned rate is negative.

The explicitly documented zero-target case =RRI(1,100,0) returns −100%. This special case should not be generalized to negative future values, which are outside the documented positive-value scope. No exact error token for those invalid inputs is asserted here.

Check the order of the starting and ending values. Reversing 100 and 81 answers a growth question from 81 to 100 rather than a decline from 100 to 81. Check that the period count represents intervals, not the number of recorded observations: annual values at the start of 2023 and start of 2026 are separated by three annual periods.

Use positive starting values and positive period counts. The source does not give every error code for text, invalid inputs, or extreme numerical values, and this guide does not import another engine’s error table. Google’s own example uses a fractional period count, so whole periods are not the only documented form, but the meaning of a fractional period still needs a consistent model.

Keep rate precision when reconstructing endpoints. Displaying 0.797414% is useful; replacing the stored monthly rate with 0.8% changes a 24-month projection.

RRI is documented in current Google Sheets. First introduction, last functioning release, and retirement are not established by the reviewed evidence. This is a rolling-service documentation review, not native validation or a financial recommendation.

Related published guides: POWER for reconstruction and ROUND for controlled display-versus-calculation choices.

First-party source reviewed October 17, 2026: Google RRI.