Abstract: The Swedish riding horse population includes about 6000 broodmares and 150 breeding stallions. The overall breeding goal is to produce riding horses competitive in several disciplines, i.e. dressage as well as showjumping and eventing. The effectiveness of this multi-purpose objective was studied in alternative strategies for selecting stallions and mares as candidate dams for them. The alternative selection strategies for stallions included 1-stage (conformation), 2-stage (conformation and performance test) and 3-stage (conformation, performance test and competition results at advanced levels) selection schemes. The mare selection schemes included the corresponding 1- and 2-stage selection schemes. The results clearly showed the 2-stage selection procedure to be most efficient, especially for stallions. It is important that the intensity in selection after the performance test is kept high. Differences in defining the breeding objective are less important, provided the selection is based on a performance test including both dressage and jumping. According to the results, the development of the Swedish performance testing scheme, in which now only 30% of the tested stallions are selected for breeding, seems justified. In mare selection schemes, performance tests also seem to be justified, especially if jumping ability is to be improved. Differences between field and station tests are only minor and since the volume of testing can be much higher in the field this would allow a much stronger selection and should thus be preferred.