Since male fertility is vitally important for cattle breeders, bull studs require means to select bulls efficiently for this trait and to cull those found below par. The present study was undertaken to investigate possible causes of variation in male fertility, measured as non-return rate (NRR), and to estimate genetic parameters for fertilizing ability, viability of the embryo, and repeatability of fertilizing ability. The material studied comprised records of single AI services, with data on two cattle breeds, Swedish Red and White and Swedish Friesian from three geographical AI areas. Each area, analysed separately, had between 40 000 and 140 000 first service records. Appreciable differences in age of cow and season of insemination were demonstrated for three different NRR, i.e. within 28, 56, and 168 days of insemination. It is suggested that these effects be taken into consideration when bulls are ranked phenotypically and when evaluated genetically for fertility, which involves examining both fertilizing ability of the sperm cells and viability of the embryo. The estimated heritabilities were 0.5–1.7% for embryo viability and slightly lower for fertilizing ability. The repeatability of a single insemination within bull was estimated to be between 0.001 and 0.006 for the different non-return rates. It is recommended that a bull's ability to fertilize is estimated in the form of a phenotypic value, real fertilizing ability (RFA). A RFA value consists of both additive genetic effects and permanent environmental effects, and should be used as the main instrument in monitoring and culling bulls for fertility. Despite a low repeatability, the correlation of a bull's true and estimated fertilizing ability reached an accuracy of 0.70 with 500 inseminations per bull. The estimated heritability for fertilizing ability is recommended for evaluation of bulls in a selection programme for improved fertility.