To estimate the effectiveness of screening for invasive cervical cancer, a case-control study was performed in Miyagi, Japan. A total of 198 cases (129 mass screen-detected and 69 outpatient-detected) of invasive cervical cancer were identified between 1984 and 1990. The results of the Papanicolaou smear of these cases were compared to those of 396 age- (+/- 5 years) and area-matched controls. Compared with women who had no prior screening (cases 51.6%, controls 16.2%), women who were screened had an odds ratio (OR) for invasive cervical cancer of 0.14 (95% confidence interval [CI] 0.088-0.230). The OR for the 175 cases of squamous cell carcinoma was 0.13 (95% CI 0.077-0.215), while the OR for the 23 cases of screened women with adenocarcinoma was 0.40 (95% CI 0.091-1.753). The time intervals following the last negative smear were assessed and we found an OR for a one-year interval and a two-year interval of 0.09 (95% CI 0.055-0.163) and 0.17 (95% CI 0.083-0.335), respectively. In conclusion, screening for invasive cervical cancer is not only effective but also the degree of protection is high for one to two years.