Introduction Although relevant, specific instruments are necessary to better understand the relationship between features of neighbourhood and health events, very few studies have developed instruments to measure neighbourhood features in developing countries.

Objective To develop valid and reliable measures of neighbourhood context useful in a Latin American urban context; assess their psychometrics and ecometrics properties and examine individual and neighbourhood-level predictors of these measures.

Methods We analysed data from a multistage (census tracts, households, and residents) household survey (2008–2009) conducted in Belo Horizonte City by the Observatory for Urban Health. One adult in each household was selected to answer the questionnaire that was composed of six domains. Neighbourhood was defined as census tract. Scales to represent different dimensions of neighbourhood were created using individual responses. Internal consistency was evaluated by Cronbach's α and three-level multilevel models were used to evaluate each scale.

Results 4048 survey respondents represented 149 census tracts. We assessed nine neighbourhood environment dimensions: Public Services (8 items), Aesthetic Quality (4 items), Walking Environment (7 items), Violence (6 items), Social Cohesion (6 items), Activities with Neighbours (11 items), Neighbourhood Physical Disorders (8 items), Neighbourhood Social Disorders (5 items) and Neighbourhood Problems (16 items). Cronbach's α coefficient ranged from 0.53 to 0.83, intraneighborhood correlation ranged 0.02–0.53 and neighbourhood reliability were 0.76–0.99. Most scales were associated with individual and neighbourhood predictors.

Conclusion These findings illustrated the measurement properties of neighbourhood-level constructs can be measured reliably, confirming their use in multilevel analysis to assess the neighbourhood effects on health outcomes.