A key problem for counterterrorism is how large numbers of individuals can be screened most efficiently to discover terrorists. This question arises at security checkpoints of all kinds, from roadblocks to airline security counters. Some argue that certain categories of individuals,
for instance, young Muslim men in the airline context, should be screened more heavily than others. Others deride this as racial profiling, and argue that any such scheme would be easily evaded. I examine a model of searching for terrorists among a population divided into categories that vary
in their potential reliability or ease of recruitment as agents of terrorist attacks. The equilibria in the model feature profiling, in that different categories are searched with different intensities. Practical difficulties in implementing a rational profiling scheme are discussed.