The recent development of powerful noise mapping softwares allows to assess the exposure to environmental noise at the scale of an agglomeration. However noise levels inside dwellings are still unknown at such scales. To evaluate indoor noise levels and to investigate the factors which determine them, an eight-day noise measurement campaign was carried out in 50 children’s home. These children were randomly chosen among 8 or 9-year-old pupils attending the public primary schools of Besançon (France). Microphones were placed in the child’s bedroom, outdoor in front of the bedroom’s facade, and in the main room. Equivalent sound pressure levels (LAeq) were calculated daily for day, evening and night periods. Standardised questionnaires were distributed to parents to describe dwelling arrangement and family habits as well as to collect noisy events occurring every day. Events were coded by periods of 30 minutes in a time-location-activity diary (TLAD). Analyses were conducted using multilevel linear regressions. A significant part of day and evening indoor noise level can be explained by non acoustic variables, especially those collected by TLAD. The results highlight the chances of improving large scale epidemiological studies based on indoor noise exposure assessment.