This paper explores the effect of certainty of punishment on different types of criminal behavior. It is argued that deterrence theory, which treats the potential criminal as weighing the relative rewards and costs of a criminal act, is more applicable to certain types of crimes than others. The strength of the negative relationship between certainty of punishment and crime rate will thus vary by type of crime. It is argued that deterrence, overload, and incapacitation theories each predict different patterns of these variations in strength. Thus the fit of the expected pattern to available data should allow us to choose among the theories. Using previously unpublished data for SMSAs and results obtained by other researchers, the actual patterns of strength of relationships are studied. Though the data presently available have many problems and some of the patterns are ambiguous, the deterrence predictions seem to have the best overall support. Interest in the deterrent effect of punishment has grown substantially in both sociology and economics in the past few years. Previously the work done by sociologists on deterrence focused on the death penalty, and these analyses indicated that the death penalty did not deter homicide (Savitz; Schuessler), so for a long time it was generally assumed by sociologists that punishment did not deter criminal behavior (Tullock). Recently, however, sociologists have expanded their analyses to include a broad range of offenses and a variety of dimensions of punishment, and they have used more sophisticated research methods. During the past few years, economists, who tend to see human behavior-including deviant behavior-in explicit rational and profit-maximizing terms, have also begun to address the deterrence question explicitly (e.g., Ehrlich; Sjoquist). Contrary to earlier results the bulk of the new evidence points to a deterrence effect (for *We would like to thank David Sjoquist for making some of his data available to us. The comments of Robert Crutchfield, Lisa Heinrich, Mayer Zald, and especially Michael Hughes on an earlier draft are greatly appreciated.