In extensive pastoral systems, where mating cannot be controlled, the breeding strategies of intermingling owner flocks interact. We present a method to evaluate a breeding programme in extensive reindeer (Rangifer tarandus) management which can also be applied to other pastoral production systems. Our main objective was to evaluate the method by applying sensitivity analyses. The method included a gene flow model with maternal effects. The potential response Rp was defined as the response that could be achieved in a closed nucleus with a given selection procedure. The value of Rp was derived from the gene flow from dams to daughters, selection differentials, and the realized difference between subpopulations. We studied a population structure having two subpopulations, only one of which had been subjected to selection. Random mating between subpopulations and a common male pool were assumed. A reference case was defined using reindeer data from the literature. After 9 years of continued selection among progeny the estimate of Rp was 7·0 times higher than the realized subpopulation difference. We analysed deviations of Rp from the reference case caused by different female age structures between subpopulations. We also analysed the sensitivity of maternal effects. The method proved insensitive to differences in female age structure between subpopulations but was sensitive to the relative contribution of maternal effects to progeny performance.