Evidence points to an association between a mother's place of residence and her newborn's health, independent of individual characteristics. Neighbourhood constructs such as immigrant density, deprivation and crime have all been separately associated with birth outcomes. Little research has considered the joint influence of variables representing a spectrum of neighbourhood constructs. Subjective vs. objective measures of neighbourhood constructs (e.g. reported vs. perceived crime) are often not considered. We sought to evaluate the relationship between neighbourhood measures of reported crime, neighbourhood perceived security, immigrant density, material/social deprivation, residential stability and the odds of small-for-gestational-age (SGA) birth in an urban setting in Canada. Neighbourhood was defined as police districts (n = 49). We linked Montreal livebirths 1997-2001 (n = 98 330) to police district crime measures, survey data on perceived security, and 2001 census data. We used multi-level analysis to calculate odds ratios (OR) for neighbourhood effects on SGA birth accounting for individual characteristics. Mothers residing in neighbourhoods with the most favourable perception had a lower odds of SGA birth than neighbourhoods with the least favourable perception [OR 0.87, 95% CI 0.77, 0.97]. Mothers in neighbourhoods with lower proportions of immigrants had lower odds of SGA birth relative to neighbourhoods with the highest proportion of immigrants. Reported crime, residential stability and material/social deprivation (accounting for neighbourhood perception) were not associated with SGA birth. Immigrant density and subjective perceptions of neighbourhood security are associated with SGA birth. Public health strategies to improve fetal growth should target neighbourhoods with low perceived security and high immigrant density.