It is generally assumed that organisms drinK ethanol during stress in order to reduce anxiety. We tested this prediction by investigating the effects of inescapable shock in relation to living conditions and the temporal pattern of ethanol consumption. Adult male rats received 60 trials of 2-sec-duration LO-mA inescapable shock each day throughout shock training. Contrary to the tension-reduction prediction, when subjects were given a choice between water and 5% ethanol, the rats living in the fearful shock environment consumed less ethanol than the rats that returned to their home cages (safety cages) after shock treatment. In addition, rats showed an in· crease in ethanol consumption, but only on the days following experience with inescapable shock. These results are inconsistent with the tension-reduction hypothesis and suggest that rats drink ethanol, not during, but when recovering from tension-arousing events.For the past quarter century, animal researchers have empirically investigated the question, "Why do organisms drink alcohol?" It has been commonly assumed that organisms drink alcohol because it reduces tension. Conger (1956) assumed that alcohol reduces tension pharmacologically and that organisms are reinforced for drinking it by tension-reducing effects. The empirical support for this "tensionreduction hypothesis" (TRH) is conflicting and paradoxical (Cappell & Herman, 1972). If ethanol is reinforcing because it reduces tension, then ethanol drinking should increase during stressful situations. Myers and Holman (1967), however, report that the stress of electric shock fails to increase ethanol drinking in rats. In contrast, other researchers report that ethanol consumption does indeed increase in stressful situations (e.g., Anisman & Waller, 1974; Cicero, Myers, & Black, 1%8;Clark & Polish, 1960;Derr & Lindblad, 1980;Masserman & Yum, 1946; Mills, Bean, & Hutcheson, 1967).Part of this confusion results from procedural differences and a failure to account for the temporal pattern of ethanol consumption. For example, Casey (1960) found that ethanol consumption increased only modestly during inescapable shock stress, but increased dramatically after shock experience ended, with peak ethanol drinking occurring 16 days after the last shock. Casey's results suggest that a necessary condition to observe an increase in alcohol self-selection is that animals have a postshock safety period. This helps us resolve an apparent contradiction between two similar experiments. Anisman and Waller (1974) found that daily periods of electric footshock (2-sec duration, 1.0-mA intensity) given every 3 min, alternated with 6-or 12-h periods of no shock, significantly increased alcohol consumption (from 2070 to 10070 ethanol solution). In contrast, Myers and Holman (1967) failed to find an increase in alcohol consumption (from 3070 to 20070 ethanol solution) when uncontrollable footshock of 2.5-sec duration and .7-mA intensity was administered one or six times per hour over a 14-day period. The key difference between ...