In its current form, the breeding program for the meat breed “Mouton Ile de France” is expected to give an annual genetic gain of 0.095 genetic standard deviation (σa) for the meat trait and 0.061 σa for the maternal trait. These results are very low compared with the dairy breeding programs modeled in this study or to the gains commonly predicted in dairy cattle. Indeed, as previously mentioned, breeding programs for small ruminants, especially meat sheep, involve inherent factors that limit genetic gains (e.g., few animals per breeding unit, limited use of AI, low capacity of progeny testing) and also have less than optimal designs. As explained above, these designs should be optimized to guarantee a fair comparison of alternative schemes. Table 3 shows that optimizing designs (current to optimized) significantly increased genetic gain in all scenarios for both meat and maternal traits. The greatest increase was observed in the reference scenario, Class-PT-culling, where the genetic gain was increased by 57.4% for the meat trait and 46.3% for the maternal trait. Optimization of design led to an improvement of genetic gain in all meat-breeding scenarios of over 20% for the meat trait and 14% for the maternal trait, except in 2 scenarios where the increase for the maternal trait was of less than 7% (i.e., GS and GS-pheno scenarios). It is also important to note that similar trends were observed by optimizing designs without modifying the rate of AI [Current to Optimized_AI (all decisional variables are optimized except the number of doses of AI); Table 3]. This suggests that genetic gain can be improved without increasing the amount of AI used in a breeding unit [i.e., by optimizing parameters such as the number of male selection candidates, male progeny tested, progeny group size, elite males, and technical weights of the selection index (wb, wm)].