Women with histopathologically confirmed cervical intraepithelial neoplasia (CIN) were followed at 3-month intervals in a randomized double-blinded trial to evaluate the efficacy of beta-carotene to cause regression of CIN. Questionnaire data, plasma levels of micronutrients, and a cervicovaginal lavage for human papillomavirus (HPV) detection were obtained at each visit, and an endpoint biopsy was performed at 9 months. Sixty-nine subjects had a biopsy endpoint evaluation; 9 of 39 (23%) subjects in the beta-carotene group versus 14 of 30 (47%) in the placebo group had regression of CIN (P = 0.039). Independent risk factors for persistent CIN at 9 months included type-specific persistent HPV infection (OR = 11.38, P = 0.006) and continual HPV infection with a high viral load (OR = 14.25, P = 0.007) at baseline and 9 months, an initial diagnosis of > or =CIN II (OR = 6.74, P = 0.016), and older age (OR for > or =25 years = 4.10, P = 0.072). After controlling for these factors, the beta-carotene and placebo groups did not differ in risk for having CIN at 9 months (OR = 1.53, P = 0.550). Resolution of baseline HPV infection was significantly correlated with non-high-risk HPV types (RR = 2.94, P = 0.015), age <25 years (RR = 2.62, P = 0.014), and douching after sexual intercourse (RR = 3.02, P = 0.012), but not with randomization group. Our data indicate that a large proportion of mild CIN lesions regress; age and HPV infection play an important role in the natural course of CIN; and repeated HPV testing may have a value in distinguishing women who need aggressive treatment for CIN versus those who do not. Supplementation of beta-carotene does not appear to have a detectable benefit in treatment of CIN.