A thorough review of the perceptual deterrence literature from 1972-1986 is provided which indicates that cross-sectional correlations between perceptions of sanction threats and self-reported criminal/deviant behavior are moderately negative for diverse offenses, consistent with the deterrence doctrine. It is noted that rather than expressing the deterrent effect, these correlations probably indicate the effect of prior behavior on currently held perceptions -- the experiential effect. In addition, since in many instances the reported correlations express simple bivariate relationships, the association may be spurious rather than causal. When researchers employing panel designs have estimated the deterrent relationship with variables in their correct temporal ordering and with more fully specified causal models, the moderate inverse effect for both perceived certainty and severity disappears. Although this would argue strongly for the continued utilization of longitudinal data and fully developed models of deterrence/social control, recent commentaries have raised questions about this line of perceptual deterrence research. These arguments are assessed, and an agenda for future deterrence research suggested. (Publisher abstract)