Following a review of previous measures and analyses of the effect of perceptions of the severity of punishment in the deterrence process, the study asserts that previous researchers have rarely tested the appropriate hypothesis with an acceptable measure of perceived severity. The study hypothesis is that perceived severity of punishment is inversely related to involvement in illegal behavior, but only among those people who perceive the certainty of arrest to be relatively high. Two measures of perceived severity are used to test the hypothesis: a refined measure, which avoids the assumption that the same punishment would be felt as equally severe by all individuals; and a conventional measure, similar to those used in previous research. Information was gathered from interviews with 400 respondents chosen at random. Respondents answered questions about self-reports of illegal behavior, perceived certainty of arrest, and perceived severity of punishment (conventional and refined measures). Analysis considered bivariate correlations and the interaction effect of perceived certainty and perceived severity. Findings suggest that perceived severity of punishment if arrested is a significant variable in the social control process, having an inverse effect on involvement in illegal behavior. That effect is concentrated among those people who believe the certainty of punishment is relatively high. Futhermore, the deterrent effect of perceived certainty of arrest varies according to the level of perceived certainty of arrest varies according to the level of perceived severity of punishment if arrested. Tabular data and footnotes are included; 43 references are provided.