The integration of routine Pap smear practice based on self-referral into an organized programme based on personal invitation is the recommended approach to cervical cancer screening. In this study, the cytology archive of the integrated cervical screening programme implemented in the Faenza district (northern Italy) was used to compare the population of women responding to invitation with that of self-referred women for the cumulative probability of early (< 36 months) repeat Pap smear after a negative result. Between 1995 and 1998, women aged 25-64 living in five municipalities of the district were targeted by the first round of the programme. Eligible for the study were 2356 women responding to invitation for screening and 2221 women not invited because of self-referral for Pap smear at any other setting. The probability of early repeat Pap smear was estimated by the life table method. Differences were evaluated by the Gehan test. Multivariate determinants were assessed by the Cox regression analysis. At 36 months' follow-up, the cumulative probability of early repeat Pap smear was 44% among self-referred women and 6% among women responding to invitation (relative hazard = 4.8). For self-referred women, the probability was related to age (with a peak at 35-44 years), previous Pap smear history and municipality of residence. Among women responding to invitation, only an inverse association with age was demonstrated. In conclusion, the observed differences in overall probability and determinants of early repeat Pap smear between the two screening populations provided important information on Pap smear usage in the total target population.