abstract A police technique known as racial profiling draws on statistical beliefs about crime rates in racial groups. Supposing that such beliefs are true, and that racial profiling is effective in fighting crime, is such profiling morally justified? Recently, Risse and Zeckhauser have explored the racial profiling of African‐Americans and argued that justification is forthcoming from a utilitarian as well as deontological point of view. Drawing on criticisms made by G. A. Cohen of the incentives argument for inequality, I argue (...) that, assuming that crime rates between different racial groups would converge in the absence of racial discrimination and unjust inequality, racial profiling is comprehensively unjustified. I also explain that, because Risse and Zeckhauser simply compare the status quo with an outcome that differs from it only in respect of the practice of racial profiling, they fail to show that such profiling is non‐comprehensively justified. And, exploring the problem from the point of view of a deontological concern for fairness, I identify a number of situations in which it would be non‐comprehensively unjustified to implement racial profiling (as well as some in which it might be non‐comprehensively unjustified not to implement racial profiling). (shrink)