To evaluate the life-prolonging effect of mass screening for breast cancer, we compared the risk of death for the patients detected by mass screening with that for the patients diagnosed in out-patient clinics, after adjusting for other relevant factors simultaneously by using the Cox regression model. A multivariate analysis using the Cox regression model in which clinical staging of disease was taken as one of the independent variables, showed that the risk of death for patients detected by mass screening was smaller by 0.765 times than that for patients found in out-patient clinics although the reduction was not statistically significant. This small reduction might be partly due to the effect of mass screening through early detection even within the same stage, and partly due to length bias, lead time bias and self-selection bias. When clinical staging of disease was removed from the independent variables, the risk of death for patients detected by mass screening was reduced from 0.765 times to 0.677 times that for patients diagnosed in out-patient clinics, which was statistically significant (P greater than 0.01). For asymptomatic patients detected by mass screening, such as reduction of the risk of death was from 0.789 times to 0.555 times that for patients found in out-patient clinics (P less than 0.05). These results suggest that mass screening for breast cancer may contribute to the reduction of the risk of death, although the effect of biases inherent in periodic screening was not removed completely in the present analysis.