Copyright © 2014 John Wiley & Sons, Ltd. Background The high rate of incarceration in the USA warrants continued exploration into understanding and ameliorating criminal behaviour. The growing use of cooperative games to measure developing prosocial behaviours has never been explored in a US criminal justice population. Aims The aim of this study is to examine cooperative game play among offenders under supervision in the community. We hypothesised that the offenders would use more guarded and self-preserving strategies and be more likely to excel in short-lived interactions than law-abiding community citizens. Methods Community supervised offenders (83) and general population comparison participants (41) were recruited by town centre adverts placed in popular shops. Using the supervision centres as venues, all participants were asked to complete four cooperative games (prisoner's dilemma, public goods game, ultimatum game and trust game), not knowing the identity of the other player who was always, in fact, the experimenter. Results The offender and general population groups were similar in age (early 30s), sex (2/3 men), race (45% white) and IQ distribution (low average range). Offenders made lower offers in the ultimatum game, had lower scores in the prisoner's dilemma, made lower investments and offered lower returns in the trust game and contributed less in the public goods game. Conclusions Even community-based offenders thus seem to have deficits in the kinds of gameplay, which are informed by theories of social cooperation, but the direction of relationship with offending remains unclear. The apparent deficits may reflect adaptation to a hostile environment where trust and reciprocity are not rewarded. It is also important to recognise that these community-based offenders did develop play indicative of trust and reciprocity, they just did so more slowly than the comparison group. This may have implications for allowing time for rapport to develop in supervisory relationships. Finally, offenders may benefit from learning that although more guarded behaviours may be adaptive in a rough neighbourhood or in jail, they may be maladaptive and limit their success in other settings such as the work place.