OBJECTIVE: Our goals were to determine the prevalence of chlamydial infection, to identify criteria for selective screening, and to compare the sensitivity of selective screening to presumptive treatment criteria in different clinical settings. STUDY DESIGN: A total of 5128 women enrolled in a cross-sectional study in public clinics in New Jersey. Univariate and multivariate methods of statistical anlaysis were used. RESULTS: The prevalence of chlamydia varied across type of public clinic and ranged from 8% to 15%. Selective screening criteria were developed for women attending each type of public clinic by use of risk factors significant in the multivariate analyses. A combination of young age and attending an urban clinic was highly predictive of chlamydia infection and identified a minimum of 85% of infected women in all public clinic settings. The use of presumptive treatment criteria identified 78% of infected women in sexually transmitted disease clinics but only 4% to 9% of infected women in other clinical settings. CONCLUSIONS: A chlamydia program that includes presumptive treatment of women attending sexually transmitted disease clinics and selective screening of women in other clinical settings where women are more likely to be asymptomatic is a clinically appropriate and economically feasible approach to directing treatment of women with chlamydial infection. (AM J OBSTET GYNECOL 1996;174:1527-33.)