By controlling for a priori risk via scores from the Massachusetts Parole Board's risk assessment form, the study concentrates on prison misconduct factors related to recidivism and not to a predetermined risk factor. Variables used to measure prison behavior included type, severity, quantity and frequency of disciplinary reports, periods spent in isolation, escape history, furloughs, returns to higher security, security from which paroled and overall movement through security levels from commitment to release. Greater statistical significance is shown for the relationship between negative prison behavior and recidivism for the 'high' risk group than is shown for the 'low/moderate' risk groups.