Background: This study tested the hypothesis that risk behaviours in disadvantaged groups would explain socio-economic inequality in dental caries prevalence among preschool children. Methods: Using a case-control study, children with caries experience (one or more decayed, missing or filled primary tooth surfaces) and with no caries experience were sampled with known probabilities from among five year olds attending the South Australian Dental Service (SADS). Dental caries experience of primary teeth was recorded by SADS clinicians. Social and behavioural information was collected using a questionnaire mailed to parents. Prevalence rates, prevalence ratios (PR) and 95 per cent confidence intervals (95% CI) were computed, taking into account sampling probabilities. Results: Questionnaires were obtained for 64.6 per cent of sampled children (n=1398) and 40.2 per cent (95% CI=37.8-42.6) of them had caries experience. Five statistically significant risk factors were identified relating to previous feeding, current oral hygiene and parent's own oral health perceptions. The prevalence of four risk factors was greater in low-income households compared with high-income households (P≤0.01). In multivariate analysis, after adjusting for age of tooth cleaning onset, age at which toothpaste was introduced was not significantly associated with caries prevalence. Behavioural risk factors did not explain incomerelated gradients in caries prevalence but modified the level of risk associated with delayed onset of tooth cleaning. Children who delayed tooth cleaning until the age of 24 months or more and who were from low-income households had a 2.7-fold increase in caries prevalence (95% CI=2.1-3.4). Conclusions: Caries prevention efforts need to target behaviours in infancy and non-behavioural risk factors among preschoolers in low-income households.