This study used data collected during the 1985 Worldwide Survey of Alcohol and Nonmedical Drug Use Among Military Personnel. The sampling units were stratified for the four service branches. In the second stage, sampling units were based on personnel rosters of the organization units selected during the first stage. The sample consisted of 25,547 active-duty personnel located at military installations in 21 countries around the world. Field survey teams contacted the personnel at their duty station, and they were requested to complete a survey questionnaire. Approximately 68 percent of personnel completed the questionnaires, yielding 17,328 observations. The scales developed describe four alcohol-use predictors: personal benefit, problem situation, peer influence, and normative definition. Multiple regression and path analysis revealed that personal benefit was the most important determinant of alcohol use. The effects of problem situations on alcohol use were largely mediated by personal benefit; moreover, peer influence not only directly influenced alcohol use but also mediated the effect of age on alcohol use. Contrary to previous research with civilian populations, normative definitions of alcohol use were not found to be a good predictor of alcohol use. 4 tables, 1 figure, 35 references, and appended descriptions of indexes and items