For the estimation of the mutation frequency, m, of induced mutations in barley, the X2 mutant frequency has been proposed as an estimator of m E(F), where E(F) is the mean segregation frequency of a homozygous mutant from a heterozygous self‐fertilized flower. Frydenberg (1963) has derived an approximation of the variance of this estimator. In the present paper a more precise estimator is given. It is demonstrated that Frydenberg's approximation overestimates the variance, and though the difference is negligible for small values of m, the use of the more precise estimator is recommended.