The effect of directly observed therapy (DOT) versus self-administered therapy (SAT) on antiretroviral (ART) adherence and virological outcomes in prison has never been assessed in a randomized, controlled trial. Prisoners were randomized to receive ART by DOT or SAT. The primary outcome was medication adherence [percent of ART doses measured by the medication event monitoring system (MEMS) and pill counts] at the end of 24 weeks. The changes in the plasma viral loads from baseline and proportion of participants virological suppressed (<400 copies/mL) at the end of 24 weeks were assessed. Sixty-six percent (90/136) of eligible prisoners declined participation. Participants in the DOT arm (n = 20) had higher viral loads than participants in the SAT (n = 23) arm (p = 0.23). Participants, with complete data at 24 weeks, were analyzed as randomized. There were no significant differences in median ART adherence between the DOT (n = 16, 99% MEMS [IQR 93.9, 100], 97.1 % pill count [IQR 95.1, 99.3]) and SAT (n = 21, 98.3 % MEMS [IQR 96.0, 100], 98.5 % pill count [95.8, 100]) arms (p = 0.82 MEMS, p = 0.40 Pill Count) at 24 weeks. Participants in the DOT arm had a greater reduction in viral load of approximately -1 log 10 copies/mL [IQR -1.75, -0.05] compared to -0.05 [IQR -0.45, 0.51] in the SAT arm (p value = 0.02) at 24 weeks. The proportion of participants achieving virological suppression in the DOT vs SAT arms was not statistically different at 24 weeks (53 % vs 32 %, p = 0.21). These findings suggest that DOT ART programs in prison settings may not offer any additional benefit on adherence than SAT programs.