The goal of captive breeding programmes is often to maintain genetic diversity until re-introductions can occur. However, due in part to changes that occur in captive populations, approximately one-third of re-introductions fail. We evaluated genetic changes in captive populations using microsatellites and mtDNA. We analysed six populations of white-footed mice that were propagated for 20 generations using two replicates of three protocols: random mating (RAN), minimizing mean kinship (MK) and selection for docility (DOC). We found that MK resulted in the slowest loss of microsatellite genetic diversity compared to RAN and DOC. However, the loss of mtDNA haplotypes was not consistent among replicate lines. We compared our empirical data to simulated data and found no evidence of selection. Our results suggest that although the effects of drift may not be fully mitigated, MK reduces the loss of alleles due to inbreeding more effectively than random mating or docility selection. Therefore, MK should be preferred for captive breeding. Furthermore, our simulations show that incorporating microsatellite data into the MK framework reduced the magnitude of drift, which may have applications in long-term or extremely genetically depauperate captive populations.