Although a significant relationship between job stress and alcohol consumption would seem quite logical, it has in fact received limited empirical support. A multivariate model is developed in this article, in which the relationship between job stress and alcohol consumption is assumed to be mediated by beliefs regarding the efficacy of alcohol to relieve stress. Considerable support for this model is demonstrated, and the implications for practice as well as research are discussed.In a recent review of the effects of alcohol and drugs on work productivity in the United States, Quayle ( 1983) estimated that approximately $30 billion a year is lost nationally due to employee alcohol and drug abuse. At least two approaches for organizations and businesses to deal with this problem have been advocated (Heneman, Schwab, Fossum, and Dyer 1986). One approach involves the use of company-sponsored occupational programs or Employee Assistance Programs (EAPs) (see Cairo [ 19831 and Weiss [ 19861 for reviews of research on EAPs; see Ford and McLaughlin [ 198 1J for a survey of corporate responses to EAPs). Although these programs vary somewhat from company to company, a common feature is that they involve formal counseling programs for treating employees with alcohol and drug problems. The strategy of an EAP approach then is to help employees overcome their substance abuse problems. Although EAPs have been widely accepted in industry (Ford and McLaughlin 1981), the actual effectiveness of these programs is difficult, if not impossible, to determine (Heneman et al. 1986; Weiss 1986).A second possible approach for dealing with alcohol abuse is to determine and ameliorate the causes of this behavior. This approach has perhaps been most succinctly stated by Baum-Baicker (1984); similar implications follow from Newman and Beehr