Differences in the prevalence of obesity between adjacent regions are quite common, but usually unexplained. This study examined whether birth place, selective migration, intelligence or education – which are both inversely and possibly causally related to obesity – are determinants of such differences. This population-based case-control study (case-cohort design) took place in the greater Copenhagen area (region 1) and surrounding provincial areas of Zealand (region 2), Denmark. A total of 2948 men with a median age of 19 years from two draft board regions during 1966–1977 were examined. The odds ratio (OR) for being obese (defined as body mass index ≥31 kg/m2) was investigated using multiple logistic regression analyses. The OR for being obese in region 2 compared with region 1 was 1·74 (1·50–2·03). Adjustment for birth place, intelligence test score and educational level reduced the OR to 1·42 (1·10–1·82). The OR for being obese for those born in region 2 compared with region 1 was 1·71 (1·46–2·01). Adjustments for intelligence test score, educational level and examination region reduced this OR to 1·13 (0·87–1·46). Irrespective of birth place, men examined in region 2 had a higher OR for being obese than those examined in region 1; this effect was most pronounced for those born in region 2 and examined in either region 1 or 2, with an OR of 1·06 (0·71–1·57) and 1·87 (1·58–2·22) respectively. In conclusion, the regional differences in the prevalence of obesity could not be explained by birth place or later selective migration, but educational level and intelligence test score did explain some of the difference.