BACKGROUND. Trachoma is still a significant problem in the developing world. Adult women are at higher risk of developing scarring and trichiasis, the potentially blinding sequelae, compared to men. In part, the higher risk may be due to more frequent infections in women because of their frequent contact with children, the main reservoir of C. trachomatis infection. However, other factors associated with infection, particularly constant infection, in adult women need to be identified. METHODS. A group of 118 women who were infected with C. trachomatis and 118 women who were not infected, but of similar age and trachoma status, were identified in 1996 from a population-based sample of women age 16 and older from eleven villages in Kongwa, Tanzania. This group of 236 was re-contacted three years later to ascertain trachoma status and determine infection status using polymerase chain reaction-enzyme immunoassay (PCR-EIA). Positive samples at both time points were examined for serovar and genotype shift, using ompA sequencing information. RESULTS. Of the original 236 women, 165 (70%) completed exams in 1999. Fifty-eight (35%) of the 165 women were excluded from this analysis because they received antibiotic treatment for trachoma in the six months prior to the second exam. Infection at baseline was the most important predictor of infection three years later (Age-adjusted odds ratio (95% confidence interval) 6.6 (1.8–24.4)). A total of 17 women (16%) were infected at the two examinations, and of the 15 for whom genotyping could be done, 11 (73%) were infected with the same ompA genotype at both time points. Chronically infected women were more likely to have trichiasis, scarring, and active trachoma at baseline than those never infected or infected only once. Only 41% of the chronically infected women were living in houses with infected pre-school children, but 24% were in houses with no children. Four of ten women with trichiasis developed incident corneal opacity over the three years. CONCLUSIONS. The data provide evidence for persistence of infection in a sub-group of women. The strongest predictor of infection at follow-up was baseline infection, and most were infected with the same genotype at both time points. For women with persistent infection, at least half were either not living with children or not living with infected children, suggesting that continual re-exposure from a close family member was less likely. Chronic infection is likely related to both exposure and immunological factors, and these need to be further identified. Inclusion of women in community-based treatment programs, regardless of whether a child is present in the house, is likely to be important in preventing the progression of inflammatory trachoma and scarring to trichiasis.