Group-to-individual inferences are inevitable. They are sometimes epistemically legitimate. Sometimes such inferences are morally odious. It is important to distinguish between morally odious and epistemically odious inferences. A prohibition of group-to-individual inferences exclusively on epistemic grounds in both unwise and impossible. These principles apply to both statistical inference and inference that is not overtly statistical. A legal case involving statistical data about the behaviour of a group of Nigerian drug smugglers is discussed.

Received on 19 December 2004. revised on 25 December 2004. accepted on 31 December 2004.