A reference horse-breeding programme with 13 500 foals each year was modelled with ZPLAN+. This new software for the optimization of the structures in breeding programmes is based on ZPLAN. In two scenarios, the implementation of a rigorous selection of mares was implemented. In scenario I, the mare performance test was the point of selection, while in scenario II, further information on 20 competitions in two more years is available. These selected mares were used for embryo transfer (ET), partly in combination with multiple ovulation (MOET). The selection intensity and the number of foals out of (MO)ET were varied in both scenarios. It was expected that 250, 500 and 1000 mares are available for selecting 20, 50, 100 or 200 donor mares each year. The number of foals out of (MO)ET was varied between one and six foals per donor mare and year. Donor mares were used for ET for 4 years. It became clear that with high selection intensities of donor mares and high reproduction rates of them, the yearly genetic gain in a horse-breeding programme could increase over a large range. In scenario II, the additional information on 20 competitions increased the accuracy of the selection index to 0.85. With 200 selected donor mares of 1000 available mares and six foals per year, the genetic gain could almost be doubled compared to the reference scenario. The implementation of ET and a related higher usage of few selected mares entails rising costs and a reduction in the genetic variance. In the most extreme MOET scenario, the effective population size was reduced by 19% relative to the reference scenario. Only if the increase in genetic gain can be converted into higher return for the breeders, the implementation of (MO)ET schemes is a realistic and sensible option for horse-breeding programmes.