As expected, increases in prison populations are usually associated with decreases in subsequent crime rates. In addition, increases in crime rates are associated with increases in subsequent prison populations. Thus, crime and prisons apparently are simultaneously determined, and instrument variables are required to separate the effects of one from the other. On a short-term basis, however, violent and property crime rates are largely independent, so there is little harm in using separate equations to determine causes of each crime type. No previous analysis of these variables has adopted this specification in its entirety. This specification strongly suggests that current crime rates and prison populations are the result of an accumulation of many small changes in precursor variables over an extended period of time. Further, there is no long-term equilibrium relationship between crime and prison, only bidirectional causality in the short term. Apart from its technical uses, this specification has implications for thinking about the problem of crime and the primary response to it, i.e., incarceration. The first implication is that a large decrease in the crime rate can only be achieved in small increments over a long period. This implication is explained, along with the implication that crime reduction from imprisonment is related only to the incapacitation of the current prison population, with no deterrent effect on those offenders still active in the community.