Previous observational epidemiologic studies of folate and cervical cancer, as well as folate supplementation trials for cervical dysplasia, have produced mixed results. We examined the relationship between serum and RBC folate and incident invasive cervical cancer in a large, multicenter, community-based case-control study. Detailed in-person interviews were conducted, and blood was drawn at least 6 mo after completion of cancer treatment from 51% of cases and 68% of controls who were interviewed. Blood folate was measured with both microbiologic and radiobinding assays. Included in the final analyses were 183 cases and 540 controls. Logistic regression was used to control for all accepted risk factors, including age, sexual behavior, smoking, oral contraceptive use, Papanicolaou smear history and human papillomavirus (HPV)-16 serology. For all four folate measures, the geometric mean in cases was lower than in controls (e.g., 11.6 vs. 13.0 nmol/L, P < 0.01 for the serum radiobinding assay). Folate measures using microbiologic and radiobinding assays were correlated (serum: r = 0.90; RBC: r = 0.77). For serum folate, multivariate-adjusted odds ratios (OR) in the lowest vs. highest quartile were 1.3 [95% confidence interval (CI) = 0.8--2.9] and 1.6 (0.9--2.9), using the microbiologic and radiobinding assays, respectively. For RBC folate, comparable OR were 1.2 (0.6--2.2) and 1.5 (0.8--2.7). Similar risks were obtained when restricting analyses to subjects with a history of HPV infection. Thus, low serum and RBC folate were each moderately, but nonsignificantly, associated with increased invasive cervical cancer risk. These findings support a role for one-carbon metabolism in the etiology of cervical cancer.