Abstract:Research is reviewed from Canada, Great Britain, and the USA, which suggests that offenders granted parole are reconvicted less than those not so released. The competing explanations for this uniformly observed difference are evaluated. It is concluded that the most likely explanation is in terms of a treatment effect rather than a consequence of the selection of better risk prisoners alone. Something about the grant or process of parole seems to work to reduce reconviction. This could be something as disreputable as the use of delaying tactics to postpone a further conviction until after licence expiry. There could also be a negative effect of parole denial to offset any positive effect of the grant of parole. Research is suggested which would clarify these points. Assuming remaining doubts are resolved in ways favourable to parole, there appears to be some grounds for optimism for parole schemes to develop into criminal justice procedures which have the net effect of crime reduction.