An objective was defined for a breeding program in South Africa to improve the female fertility of the national herd. The aim of the study was to change the approach of defining a trait based on female fertility and to define the trait, reproduction merit, as an indicator of the efficiency of male animals as sires in a given population. Reproductive merit reflects the retention of male animals in a given population, but has the added advantage that it will also yield information on the reproduction performance of a sire’s female offspring in that population. Data was extracted from the INTERGIS (Integrated Registration and Genetic Information System) for the Afrikaner beef breed in South Africa. Only data of females of sires older than nine years were included in the data to estimate the variance components, in order to give all sires an equal opportunity to have female offspring presented in all five categories. Effects included in the models were a breeder*year of birth*season (h*y*s) concatenation. For the linear analysis h*y*s was included as either fixed or random. Sire variances, using non-linear and linear models, were estimated as 0.06 (GFCAT), 0.052 (REML, h*y*s fixed) and 0.063 (REML, h*y*s random). The correlation between the solutions for the two REML models (fixed or random) was 0.73. Correlations between solutions from GFCAT and REML with h*y*s fitted as fixed or random were 0.99 and 0.73 respectively.