Objectives Living in an urban area influences obesity. However, little is known about whether this relationship is truly independent of, or merely mediated through, the demographic, socio-economic and lifestyle characteristics of urban populations. We aimed to identify and quantify the magnitude of this relationship in a Sri Lankan population.

Methods Cross-sectional study of adults aged 20–64 years representing the urban (n = 770) and rural (n = 630) populations, in the district of Colombo in 2004. Obesity was measured as a continuous variable using body mass index (BMI). Demographic, socio-economic and lifestyle factors were assessed. Gender-specific multivariable regression models were developed to quantify the independent effect of urban/ rural living and other variables on increased BMI.

Results The BMI (mean; 95% confidence interval) differed significantly between urban (men: 23.3; 22.8–23.8; women: 24.2; 23.7–24.7) and rural (men: 22.3; 21.9–22.7; women: 23.2; 22.7–23.7) sectors (P < 0.01). The observed association remained stable independently of all other variables in the regression models among both men (coefficient = 0.64) and women (coefficient = 0.95). These coefficients equated to 2.2 kg weight for the average man and 1.7 kg for the average woman. Other independent associations of BMI were with income (coefficient = 1.74), marital status (1.48), meal size (1.53) and religion (1.20) among men, and with age (0.87), marital status (2.25) and physical activity (0.96) among women.

Conclusions Urban living is associated with obesity independently of most other demographic, socio-economic and lifestyle characteristics of the population. Targeting urban populations may be useful for consideration when developing strategies to reduce the prevalence of obesity.