Severity estimates of 4 different kinds and 27 different magnitudes were obtained from a French-speaking, household-stratified sample of 299 Canadians. Results indicate that what a punishment is worth is defined by its perceived severity and grounded in social life itself. The public's views of such matters are not unreliable: their severity scales are internally consistent and show an understanding of the structural properties of the various sentencing options available to the criminal courts. Given this internal validity, penalty severity scales, if based on magnitude estimation techniques, provide a reasonable, if tentative, basis for calculating current exchange rates between qualitatively incommensurate penalties such as probation, community service, fine, or incarceration. Such rates can, in turn, be used to assess jurisdictional sentencing disparities in a rather straightforward manner and can provide some insight into sentencing reform outcomes. However, penal metric theory has yet to organize itself systematically. Sanction severity ratings appear to be status dependent; and as demands for punishment increase, perceived severity estimates decrease. Further, severity scales of special interest groups such as police or prosecutors appear to differ significantly from those of the general public. Further, the issue of whether existing scales of offense seriousness and sanction severity provide a basis for sentencing under a just-desserts logic must be addressed. 4 tables and 25 references. (Author abstract modified)