Naltrexone is a theoretically promising alternative to agonist substitution treatment for opioid dependence, but its effectiveness has been severely limited by poor adherence. This study examined, in an independent sample, a previously observed association between moderate cannabis use and improved retention in naltrexone treatment. Opioid dependent patients (N = 63), admitted for inpatient detoxification and induction onto oral naltrexone, and randomized into a six-month trial of intensive behavioral therapy (Behavioral Naltrexone Therapy) versus a control behavioral therapy (Compliance Enhancement), were classified into three levels of cannabis use during treatment based on biweekly urine toxicology: abstinent (0% cannabis positive urine samples); intermittent use (1% to 79% cannabis positive samples); and consistent use (80% or greater cannabis positive samples). Intermittent cannabis users showed superior retention in naltrexone treatment (median days retained = 133; mean = 112.8, SE = 17.5), compared to abstinent (median = 35; mean = 47.3, SE = 9.2) or consistent users (median = 35; mean = 68.3, SE = 14.1) (log rank = 12.2, df = 2, p = .002). The effect remained significant in a Cox model after adjustment for baseline level of heroin use and during treatment level of cocaine use. Intermittent cannabis use was also associated with greater adherence to naltrexone pill-taking. Treatment interacted with cannabis use level, such that intensive behavioral therapy appeared to moderate the adverse prognosis in the consistent cannabis use group. The association between moderate cannabis use and improved retention on naltrexone treatment was replicated. Experimental studies are needed to directly test the hypothesis that cannabinoid agonists exert a beneficial pharmacological effect on naltrexone maintenance and to understand the mechanism.