Objective: To investigate the association between treatment of maternal trichomoniasis with oral metronidazole and preterm birth. Methods: We conducted a retrospective cohort study using Medicaid billing data and birth certificate records for 144,737 live births in South Carolina from 1996 through 2002. Cases of trichomoniasis were identified using billing codes. The risk of preterm birth among women with treated vs. untreated trichomoniasis was modeled as a time-varying covariate in Cox proportional hazards regression. Women who were initially diagnosed with trichomoniasis at ≥35 weeks or delivered <2 days after the diagnosis were excluded from the modeling. We also examined the effect of excluding women who gave birth within 14 days following the diagnosis. Results: There were 4274 (3.0%) women diagnosed with trichomoniasis; 3579 women were diagnosed with trichomoniasis prior to 35 weeks gestation and remained pregnant until at least the second day after the diagnosis, and 1436 (40.1%) of these women filled a prescription for oral metronidazole within 14 days. Of treated women, 12.7% delivered prior to 37 weeks compared with 15.3% of women who were not treated within 14 days. In the multivariable proportional hazards regression, treatment appeared to be protective both for women with and without another genitourinary infection at some point during pregnancy (HR=0.69, 95% CI 0.50, 0.95 in women with another infection; HR=0.69, 95% CI 0.52, 0.92 in women without another infection). After excluding women who delivered within 14 days of the diagnosis, the hazard ratios remained <1.0 but were no longer statistically significant (HR=0.84, 95% CI 0.64, 1.11 in women with another infection; HR=0.96, 95% CI 0.69, 1.32 in women without another infection). Conclusions: Treatment with oral metronidazole was not associated with increased risk of preterm birth in women diagnosed with trichomoniasis.