Abstract: This paper examines the causal effect of fines and court fees on criminal reoffending for low-level misdemeanants by leveraging the quasi-random assignment of judges in North Carolina. I find strong deterrent effects, with the imposition of any fines or fees reducing the likelihood of reoffending by 9 percentage points within two years of the original offense. Treatment effects are weakly negative for almost all defendant types, contradicting the hypothesis that financial obligations induce poorer defendants to engage in crimes entailing economic gains, although I do find evidence of increased financial distress for poorer defendants. Reductions in recidivism are driven almost entirely by defendants living in wealthier neighborhoods and first time defendants. Using generic machine learning methods to fully characterize treatment effect heterogeneity, I compare alternative allocations of fines and court fees and find significant improvements, reducing reoffending by 20%, when prioritizing fines and fees to wealthier defendants.