MRSA and extensively drug-resistant TB are just two examples where we’re running out of solutions to combat pathogens. In animal health, refugia are increasingly discussed as a management strategy. Refugia are untreated subpopulations – they provide a safe haven for pathogens. By connecting refugia with drug-treated populations there is hope to limit the spread of drug resistance at an acceptable cost in terms of disease burden. Basically, the connections let drug-susceptible pathogens into the treated population to stop their drug-resistant counterparts taking over. We built a model to try and clarify when this sort of strategy might work (Park et al. 2015, Biol. Lett.). We found that the anticipated outcome (strong connections = high prevalence and low resistance in treated groups) was too simplistic. Rather, there are epidemiological and evolutionary interactions at work and these can be understood by decoupling transmission and selection through mathematical analysis.