OBJECTIVES--To identify predictors for Chlamydia trachomatis infection among visitors of an STD clinic in Amsterdam in 1986-1988. To design predictor-based screening programmes for C trachomatis. To evaluate the chosen screening strategy in 1993. METHODS--In 1986-1988, 947 heterosexual men and 648 women participated in the study. A medical history was recorded, a physical examination took place and samples were taken for laboratory diagnostics. Information for 1993 was available from routine databases. RESULTS--C trachomatis infections in heterosexual men in 1986-1988 (prevalence 15.8%) were independently associated with: age under 26 years, being an STD contact, coitarche below 13, last sexual contact with a non-prostitute, (muco)-purulent urethral discharge and ten or more leukocytes per microscopic field of urethral smear or urine. For women (prevalence 21.5%) independent predictors were age under 26, no history of STD, being an STD contact, cervical friability, (muco)purulent cervical discharge, presence of clue cells and ten or more leukocytes per field of urethral smear. Screening men with one anamnestic predictor for C trachomatis and ten or more leukocytes in smear or urine (59% of men) would detect 93% of the cases in 1986-1988. For females only universal screening proved to be suitable. After the introduction of a screening strategy in 1989 (universal for women, indicated by urethritis for men), a strong decline was found in the C trachomatis prevalence for all subgroups in 1993, excluding prostitute's clients and Turkish men. CONCLUSIONS--In 1993 the overall C trachomatis prevalence had declined among the attendants of the STD clinic. It seems likely that this fall was caused both by the screening programme and the reduction of risk behaviour.