In order to evaluate the effectiveness of a mass screening program for stomach cancer, a case-control study was conducted in Nose town in Osaka, Japan. The case series consisted of all deaths from stomach cancer during the period 1969-1981 (54 in males and 37 in females). For each case, 3 controls of the same sex and from the same precinct as the case, and born within 5 years of the case birth-year, were selected at random from Nose town residents alive at the date of death of the relevant case. We then investigated whether each case and corresponding controls had ever had screening tests before the date of diagnosis of the case. From the matched analysis of the distribution of screening in case-control combination, the odds ratio of screened vs. unscreened among those who died from stomach cancer compared to those who did not was calculated as 0.595 (90% confidence interval: 0.338-1.045) among males and 0.382 (0.185-0.785) among females. When the screening tests conducted within 12 months of diagnosis were ignored on the presumption that they were symptom-related, the odds ratio was calculated as 0.519 (0.297-0.905) among males and 0.486 (0.239-0.986) among females. These data strongly suggest that the mass screening program is effective in reducing stomach cancer mortality.