Objectives <p>This study examines whether gradual release measures, specifically open prisons and unsupervised short leave, reduce recidivism among former inmates.</p> Methods <p>Data from 907 released prisoners were analyzed. Conditional inference forests and statistical balancing methods were used to control for confounding variables, and survival models estimated effects on general recidivism and new prison sentences.</p> Results <p>Open prisons were associated with substantially lower risks of both general recidivism (<i>HRs</i> ≈ 0.66–0.72) and new prison sentences (<i>HRs</i> ≈ 0.48–0.52). Evidence for the effect of unsupervised short leave was less consistent, though analyses indicated a reduced likelihood of general recidivism for eligible inmates (<i>HR</i>_<i>ATT</i> = 0.72).</p> Conclusions <p>Findings suggest that open prisons are an effective rehabilitation measure, whereas the impact of unsupervised short leave remains uncertain. Results underscore the value of careful confounding adjustment and indicate that gradual release programs can support reintegration and reduce reoffending.</p>