Individual bacterial cells can communicate via quorum sensing, cooperate to harvest nutrients from their environment, form multicellular biofilms, compete over resources and even kill one another. When the environment that bacteria inhabit is an animal host, these social behaviours mediate virulence. Over the last decade, much attention has focussed on the ecology, evolution and pathology of bacterial cooperation, and the possibility that it could be exploited or destabilised to treat infections. But how far can we really extrapolate from theoretical predictions and laboratory experiments to make inferences about ‘cooperative’ behaviours in hosts and reservoirs? To determine the likely importance and evolution of cooperation ‘in the wild’, several questions must be addressed. A recent paper that reports the dynamics of bacterial cooperation and virulence in a field experiment provides an excellent nucleus for bringing together key empirical and theoretical results which help us to frame – if not completely to answer – these questions.