This registered clinical trial sought to validate a laboratory test system devised to screen medications for alcoholism treatment (TESMA) under different contingencies of alcohol reinforcement. Forty-six nondependent, but at least medium-risk drinkers were given the opportunity to earn intravenous infusions of ethanol, or saline, as rewards for work in a progressive-ratio paradigm. Work demand pattern and alcohol exposure dynamics were devised to achieve a gradual shift from low-demand work for alcohol (WFA) permitting quickly increasing breath alcohol concentrations (BrAC) to high-demand WFA, which could only decelerate an inevitable decrease of the previously earned BrAC. Thereby, the reward contingency changed, modeling different drinking motivations. The experiment was repeated after at least 7 days of randomized, double-blinded treatment with naltrexone, escalated to 50 mg/d, or placebo. Subjects treated with naltrexone reduced their cumulative WFA (cWFA) slightly more than participants receiving placebo. This difference was not statistically significant in the preplanned analysis of the entire 150 min of self-administration, i.e., our primary endpoint (p = 0.471, Cohen’s d = 0.215). Naltrexone serum levels correlated with change in cWFA (r = −0.53; p = 0.014). Separate exploratory analyses revealed that naltrexone significantly reduced WFA during the first, but not the second half of the experiment (Cohen’s d = 0.643 and 0.14, respectively). Phase-dependent associations of WFA with changes in subjective stimulation, wellbeing and desire for alcohol suggested that the predominant reinforcement of WFA was positive during the first phase only, and might have been negative during the second. We conclude that the TESMA is a safe and practical method. It bears the potential to quickly and efficiently screen new drugs for their efficacy to attenuate positively reinforced alcohol consumption. It possibly also provides a condition of negative reinforcement, and for the first time provides experimental evidence suggesting that naltrexone’s effect might depend on reward contingency.