A condition for effective cervical cancer screening is a low incidence of cervical cancer after negative screening compared to that in the absence of screening. This relative risk was studied for the period 1994 -1997 in the Netherlands and compared with previous studies. All cases of invasive cervical cancer diagnosed from 1994 to 1997 in the Netherlands were related to woman-years at risk, stratified by age, number of preceding negative screenings and time since the preceding negative screening. These incidence rates were compared with that before screening started in the Netherlands. The relative risk increases from 0.13 in the first year after screening to 0.24 after more than 6 years after screening for women with one previous negative screening. These figures reduce to 0.06 and 0.18, respectively, for women with two or more previous screenings. However, these estimates are less favourable when account is taken of the likely decrease in risk for cervical cancer in the period studied. Our data show a low relative risk of cervical cancer for several years following the last negative Pap smear. However, the denominator of the relative risk, that is, the incidence without screening, may have been overestimated. This applies also to the IARC multicountry study, and may have caused too optimistic expectations about the effectiveness of cervical cancer screening.