Rules restricting resource use are ubiquitous to conservation. Recent increases in poaching of iconic species such as African elephant and rhino have triggered high-profile interest in enforcement. Previous studies have used economic models to explore how the probability and severity of sanctions influence poacher-behaviour. Yet despite evidence that compliance can be substantial when the threat of state-imposed sanctions is low and profits high, few have explored other factors deterring rule-breaking. We use the randomised response technique (RRT) and direct questions to estimate the proportion of rural residents in north-western Taiwan illegally killing wildlife. We then model how potential sources of deterrence: perceived probabilities of detection and punishment, social norms and self-imposed guilt, relate to non-compliant behaviour (reported via RRT). The perceived likelihood of being punished and two types of social norms (injunctive and descriptive) predict behaviour and deter rule-breaking. Harnessing social norms that encourage compliance offers potential for reducing the persecution of threatened species.