BACKGROUND: With an improved compliance with screening a larger reduction of cervical cancer incidence would be within reach. We aimed at investigating why certain women do not attend Pap smear screening and at validating the reliability of self-reported screening. METHODS: In 1998 in the county of Uppsala, Sweden, information was collected through telephone interviews with 430 nonattendees and 514 attendees to Pap smear screening, who were all sampled from a population-based database. The women's recall of attendance was validated against the database. The main outcome measures used were odds ratios (OR) and 95% confidence intervals (CI). RESULTS: Non-attendance was positively associated with nonuse of oral contraceptives (OR = 3.56, 95% CI 2.18-5.83), seeing different gynecologists (OR = 1.90, 95% CI 1.34-2.70), and seeing a physician very often (OR = 3.12, 95% CI 1.45-6.70) or not at all (OR = 1.78, 95% CI 1.09-2.90). Frequent condom use (OR = 1.88, 95% CI 1.02-3.47), living in rural/semirural areas (OR = 1.55, 95% CI 1.07-2.21), and not knowing the recommended screening interval (OR = 2.16, 95% CI 1.20-3.89) were all associated with nonattendance, whereas socioeconomic status was not, when tested in a multivariate model. Among the nonattendees, 57% underestimated the time lapse since last smear. CONCLUSIONS: Seeing a gynecologist on a regular basis and information guiding women to have a Pap smear on their own initiative are important factors for recurrent screening. Therefore, information should be given to all women about the purpose and benefits of Pap smear testing. Self-reports on screening should be treated with caution.