The economic evidence for deterrence, as distinguished from incapacitation, is small and beset by difficulties. It is argued that criminologists would benefit from substantially scaling back the enterprise of trying to estimate the various behavioral parameters central to a micro-level approach to measuring incapacitation effects, including the annual rate of offending outside of prison and the lengths of criminal careers. However, a key question is a matter of which approach holds the greater promise for identifying policy-relevant information about incarceration and about incarceration’s effects. The most promising way to identify the net impacts of incarceration from the combination of incapacitation, deterrence, and replacement effects is from the careful study of natural experiments. Credible natural experiments require “shoe leather” research to identify cases where State or Federal policy changes, court decisions, or treatment assignment rules generate differences in punishment experiences across populations that are plausibly unrelated to other determinants of criminal activity. References