A prospective follow-up study, without further treatment, of a series of 528 women with cervical human papillomavirus (HPV) infections was conducted from 1981 to the present, with a mean follow-up time of 60.3 months (standard deviation 25.1 months). The women visited the Outpatient Department of Gynecology, Kuopio University Hospital, Kuopio, Finland at six-month intervals. At each visit, a thorough gynecologic examination, PAP-smear, and colposcopy with or without punch biopsy were performed. Epidemiologic data were collected by questionnaire, and complete follow-up data were available for 480 of the 528 women. Of these 480 cases of HPV infection, 58.3% regressed spontaneously, and clinical progression was detected in 14.8%. To establish the prognostic factors associated with the clinical course of cervical HPV infections, the Cox proportional hazards regression model was applied. In the analysis, five variables were included: age, PAP-smear class, grade of cervical intraepithelial neoplasia (CIN), HPV type, and colposcopic appearance at the first visit. In general, patient age at the time of diagnosis was inversely related to the probability of spontaneous regression (P less than 0.01). CIN II, CIN III and HPV type 16 were the most significant independent prognostic factors for progression of cervical HPV infections (P less than 0.001, P less than 0.0001, and P less than 0.001, respectively). We conclude that whenever HPV 16 DNA is found in the cervical biopsy with any grade of CIN, the lesion should be treated. Similarly, the presence of CIN II and CIN III indicates treatment whether HPV DNA are detected or not.(ABSTRACT TRUNCATED AT 250 WORDS)