The amount of help given to Blacks versus Whites is often assumed to reflect underlying levels of racism (or lack thereof). This meta-analysis assessed discrimination against Blacks in helping studies. The overall effect size for the 48 hypothesis tests did not show universal discrimination against Blacks (d = .03, p = .103). However, consistent with the predictions of aversive racism, discrimination against Blacks was more likely when participants could rationalize decisions not to help with reasons having nothing to do with race. Specifically, when helping was lengthier, riskier, more difficult, more effortful, and when potential helpers were further away from targets, less help was given to Blacks than to Whites. Interestingly, discrimination against Blacks was shown when there were higher levels of emergency. This suggests that discrimination may occur when the ability to control prejudicial responding is inhibited, or when the arousal of the emergency is misattributed to intergroup anxiety.