The beneficial effect of cervical cytology in reducing the incidence of invasive cervical cancer is well accepted, but many issues regarding specific patterns of screening remain to be resolved, and preventive programmes need to be adapted to regional characteristics. In a case-control study conducted in Latin America, we investigated cytological screening histories of 759 cases of invasive cervical cancer and 1430 controls, with participation rates of 99% and 96%, respectively. Fifty per cent of the cases and 29% of the controls reported never having been screened. Screening was less common among older, less educated and less parous women; non-users of oral contraceptives and women without histories of venereal diseases. There was also evidence that older women and those with multiple partners had longer intervals between examinations. The relative risk (RR) associated with no prior screening was approximately 3 and was not modified by other risk factors. Women reporting a Pap smear within 24-47 months before interview had the same RR as those examined within 12-23 months. Women tested longer ago had higher risks, but still much lower than women never examined. There was evidence that one examination is associated with less reduction in risk than two, regardless of the interval since last Pap smear. Screening appeared to reduce risk of both squamous cell carcinomas and adenocarcinomas. As expected, cases presenting at advanced stages were less likely to have been screened and reported longer intervals since their last examination. These results support the need to concentrate limited resources in the groups that need screening most, mainly older and less educated women who have never been screened.