Background and Objectives: Detection of subclinical Chlamydia trachomatis infection in women is a high but costly public health priority. Goals: To develop and test simple selective screening criteria for chlamydia in women, to assess the contribution of cervicitis to screening criteria, and to evaluate cost-effectiveness of selective versus universal screening. Study Design: Cross-sectional study and cost-effectiveness analysis of 11, 141 family planning (FP) and 19, 884 sexually transmitted diseases (STD) female clients in Washington, Oregon, Alaska, and Idaho who were universally tested for chlamydia using cell culture, direct fluorescent antibody, enzyme immunoassay, or DNA probe. Results: Prevalence of cervical chlamydial infection was 6.6%. Age younger than 20 years, signs of cervicitis, and report of new sex partner, two or more partners, or symptomatic partner were independent predictors of infection. Selective screening criteria consisting of age 20 years or younger or any partner-related risk detected 74% of infections in FP clients and 94% in STD clients, and required testing 53% of FP and 77% of STD clients. Including cervicitis in the screening criteria did not substantially improve their performance. Universal screening was more cost-effective than selective screening at chlamydia prevalences greater than 3.1% in FP clients and greater than 7% in STD clients. Conclusions: Age and behavioral history are as sensitive in predicting chlamydial infection as criteria that include cervicitis. Cost-effectiveness of selective screening is strongly influenced by the criteria's sensitivity in predicting infection, which was significantly higher in STD clients. At the chlamydia prevalences in the populations studied, it would be cost saving to screen universally in FP clinics and selectively in STD clinics, the reverse of current practice in many locales.