For daughter groups of 15 test bulls, controls of paternity were performed by using blood group factors and biochemical polymorphisms. Data of incorrectly assigned daughters influenced the estimation of breeding values, heritabilities and correlations for milk performance traits. Formulae are given that show the effects of variable misidentification rates on estimation of breeding values, selection intensities, heritabilities, and genetic gains. For example, for milk fat yield, the genetic gains drop at a misidentification rate of 15% between 8.7% (for h2 = .5) and 16.9% (for h2 = .2) below values attained without misidentifications. Consequently, decreasing misidentification rates in progeny of test bulls can be used to diminish the progeny size per test bull for constant genetic gain, to achieve more precise ranking of all or distinct test bulls according to their "true" breeding values and(or) to increase the number of test bulls by using the same amount of test inseminations and the same precision of ranking. Actions to reduce misidentification rates in cattle populations are discussed.