Scientific and lay theories propose that negative affect plays a causal role in problematic alcohol use. Despite this common belief, supporting experimental evidence has been mixed. Thus, the goals of this study were to (a) meta-analytically quantify the degree to which experimentally manipulated negative affect influenced alcohol use and craving in the laboratory, (b) examine whether the size of this effect depended on key manipulation characteristics (i.e., self-relevance of the stressor, timing of the end of the stressor, and strength of negative affect induction) or sample characteristics (i.e., substance use history). Across 41 studies (N = 2,403), we found small-to-medium effects for more use (dav = .31, 95% confidence interval; CI [.11, .50]) and craving (dav = .39, 95% CI [.04, .74]) following a negative affect manipulation than a control manipulation. We also found a significant increase in craving from pre- to postaffect induction (dav = .36, 95% CI [.14, .58]). This suggests the mixed results from the prior literature were likely because of statistically underpowered studies. The moderator hypotheses received weak support, with few significant results in the predicted direction. Our meta-analysis provides clarity about a previously inconclusive set of results and highlights the need for more ecologically valid manipulations of affect in future work.