Objective: To evaluate the feasibility and effects of a 12-week intervention combining aerobic and strength exercises in prison inmates with psychiatric disorders. Design: Two parallel-group, randomized controlled trials. Setting: A psychiatric prison hospital. Subjects: Forty-one men prison inmates (mean age ± SD = 38.2 ± 9.2 years, mean prison duration ± SD = 2.6 ± 2.5 years) with psychiatric disorders (primarily personality disorder, n = 27; mean illness duration ± SD = 12.0 ± 10.5 years). Interventions: Participants were randomly allocated to intervention group consisted of exercise plus usual care ( n = 21) or control group which received usual care ( n = 20) for 12 weeks. The exercise programme included three weekly sessions of group-based moderate-to-high intensity combined exercises designed and supervised by exercise professionals. Main measures: Fitness and anthropometric measures were assessed using field-based tests (6-minute walk, Incremental Shuttle Walk, Arm-Curl, and Chair-Stand), handgrip dynamometry, bioelectrical impedance, and waist and hip circumferences. Results: There were no adverse events, and 10 intervention participants withdrew. The remaining 11 participants attended a mean of 28 sessions, of which nine met the compliance criteria. Between-group change differences substantially favoured the compliance intervention group for the 6-minute walk (+21.2%), Incremental Shuttle Walk (+33.9%), Arm-Curl (+13.8%), waist (−3.5%), waist/height0.5 (−1.7%) (−2.7%), waist/hip (−3.4%), and Body Shape Index (−3.3%) (−3.5%). Additional analysis showed beneficial effects of exercise participation on handgrip strength. Conclusion: The intervention was safe, had a high dropout rate, and seemed to be effective for improving fitness and anthropometric measures in men prison inmates with psychiatric disorders who attended and participated in the exercise sessions.