1. Substance abusers subtypes have been identified considering underlying psychobiological disorder, familial factors, age of onset, legal problems and drug of choice. 2. In the present study the authors submitted 98 male heroin addicted individuals (age 19-28 y) to the Buss Durkee Hostility Inventory (Italian version) and a structured interview concerning social and clinical history; legal problems, age of onset of drug abuse, drug of choice. 3. Serotonergic system sensitivity was evaluated with fenfluramine challenge for PRL assay. 4. Thirty two patients (group A) showed high score for resentment and guilt at BDHI (hostility in), low rate of legal problems, late age of onset, preference for heroin and alcohol. Twenty nine patients (group B) showed high score for assault and irritability at BDHI (hostility out), high rate of legal problems, early age of onset, preference for heroin and cocaine. The other 37 patients (group C) showed aggression score in the normal range at BDHI, no legal problems, late onset of substance abuse, preference for heroin only. 5. PRL responses was blunted in group A (p < 0.001) and significantly decreased in group B (p < 0.05). PRL plasma levels were inversely correlated with HRSD scores. 6. All the patients were included in a treatment protocol with fluoxetine and naltrexone or placebo and naltrexone for 6 months. 7. The treatment was effective in group A with a significant improvement of BDHI results and decrease of craving score, lower level of drop out, lower level of positive urine controls. No significant differences between fluoxetine and placebo effects have been evidenced in patients of group B and C. The present findings suggest that psychopharmacological approach to addiction needs a diagnostic screening for specific subtypes.