Detection of Chlamydia trachomatis eye infection is largely unsatisfactory by standard laboratory methods. A polymerase chain reaction/enzyme immunoassay (PCR-EIA) that had previously been successful for diagnosis of genital C trachomatis infection was compared with direct antibody immunofluorescence (DFA) for detection of the organism in conjunctival scrapes. 234 Tanzanian children aged 1-7 years living in a village that had had no previous trachoma control programme were classified clinically as having no sign of trachoma (0) n = 97, follicular trachoma (TF) n = 100, or intense inflammatory trachoma with or without TF (TI +/- TF) n = 37. PCR-EIA detected C trachomatis in 24%, 54%, and 95% of subjects, respectively, compared with elementary body (EB) detection by DFA of 1%, 28%, and 60%, respectively. Overall prevalence of chlamydial eye infection was 22% by DFA compared with 48% by PCR-EIA. Of subjects with chlamydial DNA at pretreatment, 103 (92%) had no detectable chlamydial DNA at the end of 4 weeks of ocular tetracycline. The findings show that PCR-EIA is likely to affect trachoma diagnosis and epidemiology because of the increased sensitivity for detection of C trachomatis in all clinical groups; the less stringent requirements for specimen collection and transport make this method suitable for field use. Moreover, the semi-quantitative aspect of PCR-EIA may be useful for monitoring a decrease in chlamydial DNA after treatment.