ABSTRACT Use of human laboratory analogues of drinking behavior can provide an efficient, cost-effective mechanistic evaluation of a medication signal on drinking, with the result of facilitating translational work in medication development. Existing human laboratory models have yet to focus on important FDA-endpoints for alcohol use disorder (AUD) medication approval or World Health Organization (WHO) Risk criteria for clinical trial investigations. To this end, we are proposing to continue to develop a novel alcohol self-administration paradigm which models both the 1) ability to resist drinking and 2) heavy drinking. To model the ‘ability to resist’ drinking, we are proposing to adapt procedures from our successful ‘ability to resist’ smoking models that we have been developing and utilizing for the past 18 years. Our smoking models have demonstrated predictive validity with regards to smoking cessation medications with large effects and have been utilized widely by the research community and the pharmaceutical industry. We are planning to develop our ‘ability to resist’ drinking models following the staged process that we have used to develop our smoking models. For this 2-year application, we are proposing to develop two versions of the ‘ability to resist’ drinking model designed to screen AUD medications which map onto clinically meaningful endpoints. Model 1 will map onto key FDA endpoints for AUD medication approval: % abstinent, and % with no heavy drinking. Model 2 will map onto non-abstinent outcomes (i.e., WHO Risk criteria). We also plan to evaluate potential mechanisms underlying the ability to resist drinking and ad-libitum drinking behavior across the two models. Positive findings from this proposal will 1) provide the necessary data to proceed with validating the models using medications with known clinical efficacy for AUD and 2) ultimately provide an important resource to the research community that will facilitate translational work in medication development for alcohol use disorder.