OBJECTIVE The geographical maldistribution of physicians is an important issue for health policy. This study aimed to evaluate recent trends in Japan and identify the factors associated with starting a rural practice. METHODS A longitudinal study was conducted using secondary analysis of national census data. The number of physicians in each region was described in 1996, 2006, and 2016 and, for two cohorts, 1996 to 2006 and 2006 to 2016; a logistic regression analysis was used to identify factors related to starting a rural practice after 10 years. RESULTS Between 1996 and 2016, the number of physicians in rural areas increased by 4%. Results of the logistic regression analysis showed that, in both cohorts, the factor with the highest odds ratio for starting a rural practice after 10 years was initial academic hospital experience. Other factors showing positive correlations were additional hospital practice and being qualified and over 30 years old, whereas female sex and aged at least 40 years had negative correlations. CONCLUSIONS From 1996 to 2016, despite the increase in the total number of physicians, geographical maldistribution increased. Some findings support the policy that some physicians should work in rural areas in return for scholarships.