Objective: This study tested the comparative effectiveness of modified behavioral self-control therapy (MBSCT) and naltrexone (NTX), as well as the added benefit of combining the 2, in problem drinking men who have sex with men (MSM) seeking to reduce but not quit drinking. Method: Participants (N = 200) were recruited and urn randomized to 1 of 2 medication conditions, NTX or placebo (PBO), and either MSBCT or no behavioral intervention, yielding 4 conditions: PBO, NTX, MSBCT, and NTX + MSBCT. In addition, all participants received a brief medication compliance intervention. Participants were treated for 12 weeks and assessed 1 week after treatment completion. Two primary outcomes--sum of standard drinks and number of heavy drinking days--and 1 secondary outcome-percentage of those drinking in a nonhazardous manner (NoH)--were selected a priori. Results: There was a significant main effect for MBSCT (all ps less than 0.01) but not NTX on all 3 outcomes. In addition, the combination of NTX and MBSCT was not more effective than either MSCBT or PBO. There was a significant interaction effect on NoH, such that NTX significantly increased the likelihood (odds ratio = 3.3) of achieving a nonhazardous drinking outcome relative to PBO. In addition, NTX was significantly more effective than PBO on a descriptive outcome: negative consequences of drinking. Conclusions: There was no advantage to adding NTX to MBSCT. In addition, MBSCT showed stronger evidence of efficacy than NTX. At the same time, NTX delivered in the context of a minimal medication compliance intervention was significantly more effective than PBO on an important clinical indicator. Results provide new information to guide the treatment of problem drinking, including in primary care settings. (Contains 2 tables and 3 figures.)