In 'Making Confinement Decisions' (Zedlewski, 1987), it is proposed that 1 year's imprisonment involves a total social cost of $25,000; whereas the social costs averted by that imprisonment through incapacitation alone are $430,000. However, an analysis of these estimates with reference to trends in the United States produces anomalous results. If the study estimates were correct, criminal justice expenditures would have decreased recently, and crime rates would have dropped toward zero as the prison population doubled. Six features of the study contribute to the wild overestimates of the incapacitative potential of expanding imprisonment: the high estimates of incapacitation effects; disregard of the diminishing marginal returns of predictive imprisonment; the problematic allocation of public expenditures such as police budgets as costs of crime; the 'average cost' treatment of all offenses (other than drug crimes) as representing equal social costs; the assumption that the costs of crime prevention and law enforcement are marginally proportionate to crime rates; and the imprecision and vagary associated with the concept of the social cost of crime and, thus, of any cost-benefit comparison. There may be many ways to measure crime costs, but even if attempts are confined to costs in dollars, the attempt to aggregate guard dog food costs with victim dollar losses and variations in police pension levels into a single monetary total seems misconceived. 1 figure and 10 references. (Author abstract modified)