Behavioral mechanisms for avoiding inbreeding are common in the natural world and are believed to have evolved as a response to the negative consequences of inbreeding. However, despite a fundamental role in fitness, we have a limited understanding of the cues that individuals use to assess inbreeding risk, as well as the extent to which individual inbreeding behavior is repeatable. We used piecewise structural equation modeling of 24 years of data to investigate the causes and consequences of within- versus extra-group paternity in banded mongooses. This cooperatively breeding mammal lives in tight-knit social groups that often contain closely related opposite-sex breeders, so inbreeding can be avoided through extra-group mating. We used molecular parentage assignments to show that, despite extra-group paternity resulting in outbred offspring, within-group inbreeding occurs frequently, with around 16% litters being moderately or highly inbred. Additionally, extra-group paternity appears to be plastic, with females mating outside of their social group according to individual proxies (age and immigration status) and societal proxies (group size and age) of within-group inbreeding risk but not in direct response to levels of within-group relatedness. While individual repeatability in extra-group paternity was relatively low, female cobreeders showed high repeatability, suggesting a strong constraint arising from the opportunities for extra-group mating. The use of extra-group paternity as an inbreeding avoidance strategy is, therefore, limited by high costs, opportunity constraints, and the limited reliability of proxies of inbreeding risk.