Fine reductions for self-reported offenses entail a potential trade-off. On the one hand, inducing offenders to self-report allows the social planner to save on enforcement costs and reduce harm through early detection. On the other hand, fine reductions may also reduce deterrence: offenders anticipate that if their detection probability turns out to be higher than initially expected, they can exploit the possibility of a more lenient sanction. We analyze how this trade-off is affected by the potential offender’s utility function, contrasting standard neoclassical preferences with loss aversion. For loss aversion, we apply the approach by Koszegi and Rabin (2006, 2007), in which reference points are determined by the ex ante expectations of equilibrium strategies. Assuming that the private benefit from crime is lost in case of detection, we distinguish between loss aversion in the fine dimension and in the benefit dimension. Intuitively, one might assume that loss aversion facilitates law enforcement because losses loom larger than gains, which sets incentives to refrain from crime. We show that a sufficient condition for this intuition to hold is that the degree of loss aversion in the fine dimension is weakly above the degree in the benefit dimension.