Cancer mortality is increasing as the population ages in Japan. Cancer registry data obtained precisely serves as the basis for planning effective cancer control measures. Three time-related factors affect cancer mortality, one of which is the birth cohort effect. Previous descriptive epidemiological studies suggest that the birth cohort effect is not negligible in cancer mortality. In this study, we developed a statistical method to automatically detect and evaluate the statistical significance birth cohort effects in cancer mortality data, using a varying coefficient model. As a result, birth cohort effects detected by the proposed method are in good agreement with previous epidemiological findings.