Two studies provided evidence that alcohol's relationship to psychological stress is indirect and is mediated by the allocation of attention. Study I found that, as the attentional demands of a distract- ing activity increased, so did alcohol's reduction of anxiety. Study 2 replicated this effect and found that a highly demanding activity could reduce anxiety even without alcohol. This study further implicated the role of attention in anxiety reduction by demonstrating a relationship between changes in anxiety and response latency to a secondary monitoring task. Finally, in both experi- ments, intoxicated subjects who did not perform any activity showed an increase in anxiety. From these data, we argue that alcohol affects psychological stress, to an important degree, through its ability, in conjunction with ongoing activity, to affect the amount of attention paid to stressful thoughts. A seminal article by John J. Conger(Conger, 1956) set forth the argument that alcohol's addictiveness stemmed in large part from its ability to reduce tension. The rewards of this "tension reduction" presumably stimulated and sustained drinking that, in time, led to addiction. It is probably not unfair to say that Conger's tension-reduction hypothesis (TRH) has guided much of the psychological research conducted on the relationship be- tween alcohol and psychological stress since its publication. Un- fortunately, a predictable relationship between alcohol and psy- chological stress has proven difficult to establish (cf., Cappell & Herman, 1972; Marlatt, 1976; Steele & Josephs, 1988). Recent research (e.g., Abrams & Wilson, 1979; Hull & Young, 1983; Levenson, Sher, Grossman, Newman, & Newlin, 1980; Polivy, Schueneman, & Carlson, 1976; Sher & Levenson, 1982; Steele & Josephs, 1988; Steele, Southwiek, & Pagano, 1986) has focused on establishing the conditions under which alcohol affects psychological stress. In this article, we seek to further elucidate these conditions and the mechanisms mediat- ing the alcohol/psychological stress relationship. We (Steele & Josephs, 1988) have demonstrated both anxio- lytic (anxiety-decreasing) and anxiogenic (anxiety-increasing) effects of alcohol intoxication. Briefly, our method is to give sub- jects alcohol or placebo drinks (all subjects are told they are