Moderate doses of alcohol impair performance on a variety of information processing tasks. Two separate meta-analyses were conducted on the results of (1) reaction time studies, and (2) recognition memory studies, representing 25 and 16 different task conditions, respectively. In both cases, performance with alcohol (either 0.8 or 1.0 ml/kg body weight) was plotted as a function of performance with no alcohol. For reaction time, a linear fit accounted for 99.7 per cent of the variance. The same function applied not only to the mean but to the distribution of reaction times from the 5th to the 95th percentiles. For recognition memory, a linear fit accounted for 96.2 per cent of the variance in accuracy (expressed as the logarithm of proportion correct). Thus alcohol appears to have a general linear effect on information processing, rather than specific effects on a subset of stages. It is concluded that the results are consistent with a reduced processing resources hypothesis for the impairment with alcohol.