The most severe forms of conduct disorder (CD) are highly stable and disabling disorders, more likely to persist in time and to evolve into disruptive or antisocial behaviors. One crucial issue in the prognosis of these forms of CD is the high resistance to both non-pharmacological and pharmacological treatments, with antipsychotic drugs being frequently used in refractory cases. Aim of this study was: (1) to explore efficacy and tolerability of olanzapine treatment in adolescents with severe CD; (2) to identify predictors of olanzapine treatment outcome. This was a retrospective study, based on clinical records of the first 23 adolescents diagnosed as having a CD, diagnosed with a clinical interview (K-SADS), either pure or with comorbid diagnoses, and treated with olanzapine. All these patients did not respond satisfactorily to non-pharmacological intervention and to adequate dosages of mood stabilizers (lithium and/or valproate). The sample consisted of 16 males and seven females, 16 inpatients and seven outpatients (mean age 13.6 +/- 1.9 years, range 11-17.2 years), followed-up for a period ranging from 6 to 12 months (mean 8.8 +/- 2.7 months). Outcome measures included the Modified Overt Aggression Scale (MOAS), Clinical Global Impression-Improvement (CGI-I) and Children Global Assessment Scale (CGAS). During the follow-up, all patients were involved in non-pharmacological treatments (psychotherapy, family therapy, or day-hospital group treatments). Based on both an improvement of at least 50% at MOAS and a score 1 or 2 at CGI-I, 14 out of 23 patients (60.9%) were classified as responders at the end of the follow-up. Significant improvement at the last observation was found in MOAS (P < 0.001) and CGAS (P < 0.001) scores. Olanzapine dosage was 8 +/- 3.2 mg/day (range 5-20 mg/day). Mean weight gain at the end of the follow-up was 4.6 +/- 3 kg. The predictors of a positive treatment response was an impulsive-affective versus controlled-predatory type of aggression. Age at onset of CD and comorbid disorders did not affect treatment response. These preliminary findings suggest that olanzapine may improve behavior in adolescents with severe and treatment-refractory CD and impulsive aggression.