OBJECTIVE: The aim of this study was to determine the applicability of the published clinical cut-off scores of the Child Behaviour Checklist (CBCL) for the classification of behaviour disorders. METHODOLOGY: Child Behaviour Checklists were obtained for 1342 subjects newly referred to the six major mental health centres in Melbourne. The normative community sample of 1002 7-, 12- and 15-year-olds was drawn from a school-based asthma prevalence study. RESULTS: The mean total problem T-score for the children referred to mental health centres was 67 and was above the clinical range for all age groups. Using referral to psychiatric services as the gold standard, the sensitivity and specificity of the CBCL using a cut-off of > or = 60, was 77.4 and 83.2%, respectively. This compares favourably with the sensitivity of 68% and specificity of 82% for the American sample. Using a cut-off score of > or = 63, the sensitivity was 70.5% and the specificity was 88.6%. The referred and community samples differed with respect to socio-economic status, family structure and mothers' level of education. Fifty-two per cent of the clinically referred children lived with both parents, compared with 89% of the community sample. CONCLUSIONS: While there are some limitations to this study in terms of both the clinic and community sample, support is provided for the usefulness and applicability of the recommended CBCL cut-off scores in an Australian population.