This paper examines an apparent contradiction in recent comparative research on crime. Most quantitative studies assume that social-structural variables which cause crime operate uniformly across different crime types — for example, income inequality increases both murder and theft rates. However, our review of this research, and our own multivariate analysis of murder and theft rates in 47 countries, suggest that social-structural variables have opposite effects on murder and theft rates. The most important predictor of both murder and theft in our analysis is population growth. Drawing on case studies and comparative essays, we offer a preliminary theoretical explanation for the opposite effects of social-structural variables — especially population growth — on murder and theft. This explanation is based on variations in the opportunities and incentives for murder and theft in developing and developed countries. We suspect that population growth measures the strength of primary group relations; that societies characterized by frequent and intense interaction between primary group members provide many opportunities for murder, but because of effective surveillance, few opportunities for theft.