Until 1975, belief in the deterrent value of capital punishment had no authoritative empirical support. Then Isaac Ehrlich reported results of a complex econometric analysis of aggregate United States time-series data, indicating that capital punishment is an effective deterrent. Considerations of earlier studies provides support for the theory that deterrent effects can be more efficiently estimated from observations of temporal differences in the relevant variables in cross-sectional data. This procedure is applied to data from the cross-section of States for 1960 and 1970. The findings, as well as earlier ones by Passell, Sellin, Schuessler, Zeisel, and others indicate that capital punishment does not have a perceptible influence on the homicide rate. Even if some people are in fact deterred from committing murder by the prospect of being executed rather than subjected to a lengthy term of imprisonment, the available evidence suggests that as many people, and perhaps more, are provoked to commit murder by the existence of the death penalty. This may be due to the increased incentives to kill police and witnesses to avoid capture for capital crimes, a decreased tendency to convict persons for capital crimes, the attraction to capital crimes by those with suicidal tendencies, or the tendency of capital punishment to increase the acceptability of violence. The apparent conflict in evidence between the findings of Ehrlich and those reported in this study may be due to Ehrlich's strong a priori belief in the theory of deterrence; the availability of a variety of small, volatile data sets; and a larger variety of options for analyzing these data sets. In sum, while the deterrence theory may apply to most sanctions, the evidence indicates that the death penalty is an important exception, one for which other theories of behavior appear to be more applicable. Tabular data and 48 footnotes are provided.