Captive breeding is the last resort to protect a species that will become extinct because it is incapable of surviving in the wild without human intervention. The freshwater pearl mussel Margaritifera margaritifera is a prime candidate for such action: the species has experienced a recent range-wide decline, particularly in Europe. In Northern Ireland, M. margaritifera is now only found in six rivers and, because of a lack of natural recruitment in the wild, it has been predicted that these populations will become extinct within 80 years. Consequently, an ex situ breeding program was established with the aim of using a semi-natural method to propagate mussels for restocking. In the present study, we analysed the levels and patterns of genetic diversity in the captive-bred mussels, as well as their parent broodstock and remnant populations from their natal and other rivers in Northern Ireland where M. margaritifera is still found, to help develop best practice for the captive breeding program. Levels of genetic diversity were high, although there was a strong correlation between genetic diversity and census population sizes, as well as evidence of inbreeding in all populations. Small but significant levels of genetic differentiation were observed between both the captive-bred juveniles and their parent broodstock, and the samples from their source river, indicating a possible founder effect. In addition, three groups of genetically distinct clusters were observed among the remaining natural populations that could be used as the basis for defining conservation units. These findings allow us to make recommendations for efficient management of the captive breeding program, including rotation of the ex situ broodstock on a regular basis to avoid prolonged inbreeding, and the establishment of captive breeding of other populations to reflect genetic differences between putative conservation units in Northern Ireland and elsewhere.