HIV testing of individuals presenting to outpatient medical clinics has generally been based upon a selection system, with testing limited to those having signs or symptoms previously found associated with HIV-1 infection among hospitalized patients. However, little is known about the efficacy of this approach, particularly in Africa. Among patients presenting to a large outpatient infectious disease clinic in Dakar, Senegal, the utility of using specific demographic and behavioral characteristics and individual presenting complaints to identify individuals with previously undiagnosed HIV-1 or HIV-2 infection was examined. Using a simple statistical approach, a composite screening rule was estimated to identify subjects with the highest probability of testing HIV positive, ie, patients who would most benefit from HIV testing. Using the presenting complaint allows identification of 83% of HIV-infected women by testing only 35% of women presenting to the clinic. Similarly, using the presenting complaint and various demographic and behavioral characteristics, it was possible to identify 84% of HIV-infected men by screening 40% of men presenting to the clinic. This study suggests that this method might provide a cost-effective approach that permits limited screening resources to be spent in a way that maximizes individual and societal benefit.