Social determinants of cancer mortality attract much attention worldwide, and several studies on statistical analysis of determinants using linear regression method (OLS) have been reported.In this study, a large-scale experiment has been done to analyze determinants of cancer mortality rates of 147 countries using 40 explanatory variables of 5 fields and a non-linear regression support vector machine (SVM) technique.Except for five developing countries, the resulting SVM model shows that just 14 factors represent cancer mortality rates of 142 countries with a higher accuracy of determination coefficient of 0.730 than previous works.It also demonstrates that the degree of national development such as intelligence quotient and health expenditures most significantly contributes to cancer mortality rates.