Most studies that link neighbourhoods to disease outcomes have represented neighbourhoods as area-level socioeconomic status. Where objective contextual attributes of urban environments have been measured, few studies of food availability have evaluated mortality as an outcome. We sought to estimate associations between the availability of fast-food restaurants (FFR), fruit and vegetable stores (FVS), and cardiovascular mortality in an urban area. Food business data were extracted from a validated commercial database containing all businesses and services in the Montréal Census Metropolitan Area (MCMA). Mortality data (1999-2003) were obtained for the MCMA (3.4 million residents). Directly standardised mortality rates for cardiovascular deaths (n = 30,388) and non-cardiovascular deaths (all causes - cardiovascular deaths) (n = 91,132) and FFR and FVS densities (n/km2) were analysed for 845 census tracts. Generalised additive models and generalised linear models were used to analyse food source-mortality relationships. FVS density was not associated with cardiovascular or non-cardiovascular mortality (relative risk (RR) = 1.02, 95% confidence interval (CI): 0.76, 1.36, and RR = 1.14, 95% CI: 0.87, 1.50, respectively). Higher FFR density was associated with mortality in bivariate and multivariable analyses. Relative risks of death (95% CI) per 10% increase in FFR density were similar for both cardiovascular and non-cardiovascular mortality: 1.39 (1.19, 1.63) and 1.36 (1.18, 1.57), respectively, accounting for sociodemographic covariates. FFR density is associated with cardiovascular mortality but this relationship is no different in magnitude than that for non-cardiovascular mortality. These results together with null associations between FVS density and mortality do not support a major role for food source availability in cardiovascular outcomes.