The purpose of this study was to present an empirical correction of the KR21 (Kuder Richardson test reliability) formula that not only yields a closer approximation to the numerical value of the KR20 without overestimation, but also simplifies computation. This correction was accomplished by introducing several correction factors to the numerator of the KR21 to reduce the inflation of the error-variance estimate. The reliability coefficients resulting from this analysis were compared under varying levels of student sample size, number of items in the test, examination mean difficulty, and examination mean discrimination. Test data were taken from a representative sample of 101 university-level classroom achievement tests. The hypothesis of no difference between the mean KR20 and KR21 coefficients was rejected, while the same hypothesis for the empirically adjusted KR21 and the KR20 was retained. The mean underestimate of the KR20 by the KR21 was 16.1 percent, while the corrected formula 21 underestimated by only 4.2 percent. The .8 correction factor was superior to all other factors and to the original KR21, as estimates of the KR20. The representativeness of the sample of tests used in this study suggested that these findings are generalizable. (Author/CP)