Relative and population attributable risks for invasive cervical cancer in different age strata relative to screening practices have been estimated using data from a case-control study conducted since 1981 in the greater Milan area, northern Italy. A total of 548 women under 75 years of age with a histologically confirmed diagnosis of invasive cervical cancer were compared with 515 controls admitted to hospital for a spectrum of nongynecological, hormonal or neoplastic, acute conditions. The percentage of never-screened women increased with age: no Pap smear was reported in 34% of controls aged 44 years or younger and 63% of older women (greater than or equal to 65 years). A similar trend emerged in recency of Pap smear, with 14% of older controls and 52% of the younger group reporting a cervical smear within 3 years before the interview. The number of Pap smears was strongly and inversely related to cervical cancer risk at all ages. Compared with no Pap smear, three or more cervical smears decreased the risk of invasive cervical cancer by about 90%. Compared with women screened the last time 6 years before the interview or never, relative risks were about 60% lower in women reporting their last Pap within 2 years in all age groups. A slightly larger proportion of older cases were attributable to the absence or delay in the screening: population attributable risk estimates were 78% in women up to 55 years of age, and 86% in those 65 years or older. Thus, deficiencies in screening were greater at ages at which the incidence of disease was higher.