Abstract Objectives The first aim was to evaluate BMI cross-sectionally over a period of 14 years (1992 to 2005) in 43 343 army men and the second was to compare BMI using the paired data of 1497 army men. The data were analysed as a function of the military ranking system, used as an indicator for socio-economic position. Design Multiple cross-sectional and longitudinal design. Results A significant increase of BMI between age categories was detected over the 14-year period; BMI remained stable in each age category. In the paired cohort, median BMI increased during the same period from 23·9 (interquartile range 3·3) kg/m 2 to 24·7 (interquartile range 3·5) kg/m 2 ( P < 0·0001). This age-dependent evolution was present in all military rankings. From age 40 years or more, BMI indicated a significant increase in the prevalence of overweight and obesity. Conclusion For the total cohort, BMI remained stable in each age category. For the paired cohort, BMI increased over time. The military leadership should emphasize prevention in order to reduce the health-care costs and disease burden in this cohort. This emphasis on prevention should target those aged less than 40 years.