Behavioral pharmacology paradigms have been used for early efficacy testing of novel compounds for alcohol use disorder (AUD). However, the degree to which early efficacy in the human laboratory predicts clinical efficacy remains unclear. To address this gap in the literature we employed a novel meta-analytic approach. We searched the literature for medications tested for AUD using both behavioral pharmacology (i.e., alcohol administration) and randomized clinical trials (RCTs). For behavioral pharmacology, we computed medication effects on alcohol-induced stimulation, sedation, and craving during the alcohol administration (k = 51 studies, 24 medications). For RCTs, we computed medication effects on any drinking and heavy drinking (k = 118 studies, 17 medications). We used medication as the unit of analysis and applied the Williamson-York bivariate weighted least squares estimation to preserve the errors in both the independent and dependent variables. Results, with correction for publication bias, revealed a significant and positive relationship between medication effects on alcohol-induced stimulation (β = 1.18 p < 0.05), sedation (β = 2.38, p < 0.05), and craving (β = 3.28, p < 0.001) in the laboratory, and drinking outcomes in RCTs, such that medications that reduced stimulation, sedation, and craving during the alcohol administration were associated with better clinical outcomes. A leave-one-out Monte Carlo analysis examined the predictive utility of these laboratory endpoints for each medication. The observed clinical effect size was within one standard deviation of the mean predicted effect size for all but three pharmacotherapies. This proof-of-concept study demonstrates that behavioral pharmacology endpoints of alcohol-induced stimulation, sedation, and craving track medication effects from the human laboratory to clinical trial outcomes. These results apply to alcohol administration phenotypes and may be especially useful to medications for which the mechanisms of action involve alterations in subjective responses to alcohol (e.g., antagonist medication). These methods and results can be applied to a host of clinical questions and can streamline the process of screening novel compounds for AUD. For instance, this approach can be used to quantify the predictive utility of cue-reactivity screening models and even preclinical models of medication development.