Background: The present study investigated the differential predictive value of parents', teachers', and clinicians' reports of psychopathology for poor outcome in children referred to a child psychiatric outpatient clinic. Method: A referred sample (N = 96), aged 6 to 12 years at initial assessment, was followed up after a mean interval of 3.2 years. Data on parent- and teacher-reported problem behavior (Child Behavior Checklist and Teacher's Report Form), and clinician-reported observations and self-reports during a semi-structured clinical interview (SCICA), were linked to outcome measures assessed with a parent questionnaire, including outpatient and inpatient treatment at Time 2, parent's wish for professional help for the child, school problems, and police/judicial contacts. Results: Information from all three informants (clinicians, parents, and teachers) predicted measures of poor outcome after three years. Clinicians' ratings on the SCICA predicted all five outcome measures. Independent of CBCL and TRF scores, SCICA scores predicted parental wish for help and inpatient treatment. Conclusions: The present study was the first to report that clinician's ratings of self-reported and observed behaviors in a semi-structured interview (SCICA) make an important unique contribution to the multiaxial assessment of problem behaviors.