Background The prevalence of overweight and obesity in the US population has risen dramatically in recent years. To try to explain this, some studies have examined the association between the built environment and obesity (measured using the body mass index (BMI)). Most of these studies have not sought to identify causal effects, but rather correlations. Methods Data from the Twin Cities Walking Study were used to examine the effect of population density and block size on BMI. Although the Twin Cities Walking Study is a cross-sectional observational study, the matched-sampling design is novel in that it maximises environmental variance while minimising person variance to enhance exchangeability of subjects and more closely mimic an experimental study. Results Contrary to expectations, the hypothesised most walkable neighbourhood (high density, small block stratum) had the greatest mean and median BMI. After adjusting for demographic covariates, physical activity and clustering due to neighbourhood, no conclusive effect of population density by block size on BMI was found (β=-1.024, 95% CI -2.408 to 0.359). Conclusion There is no evidence of an effect of population density by block size on BMI.