The authors used cell cultures to assess the prevalence of cervical infection caused by Chlamydia trachomatis in a population-based sample of nonpregnant women aged 15 to 34 years who attended two primary care clinics at Group Health Cooperative of Puget Sound, Seattle, Washington, between January 1, 1988, and June 30, 1989. C. trachomatis was isolated from 67 of 1,804 women (3.7%), including 13% of those who were less than 20 years of age. Seven patient characteristics were independently predictive of chlamydial infection by stepwise multivariate logistic regression analysis: being unmarried, examination showing cervical ectopy, black race, douching, nulliparity, age of 24 years or less, and intercourse with two or more partners within the preceding year. Testing all women who had a score of 5 or more (28% of women) on a weighted index based on risk factors would detect 77% of all infections with a positive predictive value of 9%. These results suggest that it should be feasible to develop a risk factor-based program to screen for cervical infection with C. trachomatis in populations where its prevalence is low.