Recent large declines in house sparrow Passer domesticus L. populations in many European towns and cities have generated much speculation as to possible environmental causes, although empirical insights have been lacking. We measured reproductive success in a declining house sparrow population along an urbanization gradient in Leicester, England, and used mixed models to identify correlates of nesting success. A stochastic simulation approach was developed to estimate annual reproductive success when the number of breeding attempts was unknown, and to assess the sensitivity of reproductive success to variation in different environmental factors. During 2 out of 3 years, reproductive output was lower than the predicted threshold required for population stability. Years of inadequate reproduction were characterized by lower chick survival and body mass at fledging (a predictor of post-fledging survival). Factors associated with reduced reproductive output included low ambient temperatures, extremes of rainfall, high levels of vegetable material in the diet (all affecting chick survival and body mass), low aphid densities (an important component of chick diet) and high concentrations of air pollution from traffic (affecting chick body mass). Reproductive failure linked to inadequate invertebrate availability provides a plausible demographic mechanism accounting for declines in urban–suburban house sparrow populations. Increased traffic volume and vegetation removal are candidate causal factors in these declines although the mechanism underlying the former is unclear. Increasing densities of key invertebrate prey away from major roads should be an objective of conservation management aimed at urban–suburban house sparrows. This study illustrates the utility of mixed models for the analysis of individual sibling survival and of stochastic simulation for modelling reproductive success in multi-brooded animals.