The purpose of this study was to evaluate the association of pharmacogenomic factors and clinical outcome in autistic children and adolescents who were treated with risperidone for long periods. Eighty-two autistic subjects diagnosed with DSM-IV and who were treated with risperidone for more than 1 year were recruited. Pharmacogenomics and clinical outcome (CGI-I, aggressive, overactivity and repetitive score) were evaluated. Almost all patients showed stable symptoms on aggressive behaviour (89.02%), overactivity (71.95%), repetitive (70.89%) behaviour and all clinical symptoms (81.71%). Only 4.48% of patients showed minimally worse CGI-I score. Patients in the non-stable symptom group had DRD2 Taq1A non-wild-type (TT and CT) frequencies higher than the clinically stable group (p = 0.04), whereas other gene polymorphisms showed no significant association. Haplotype ACCTCAT (rs6311, rs1045642, rs1128503, rs1800497, rs4436578, rs1799978, rs6280) showed a significant association with non-stable clinical outcome (χ = 6.642, p = 0.010). Risperidone levels showed no association with any clinical outcome. On the other hand, risperidone dose, 9-OH risperidone levels and prolactin levels were significantly higher in the non-stable compared to the stable symptom group (p = 0.013, p = 0.044, p = 0.030). Increased appetite was the most common adverse drug reaction and associated with higher body-weight, whereas it was not significantly associated with genetic variations and non-genetic information. In conclusion, risperidone showed efficacy to control autism, especially aggressive symptoms in long-term treatment. However, Taq1A T - carrier of dopamine 2 receptor gene - is associated with non-stable response in risperidone-treated patients. This study supports pharmacogenomics testing for personalized therapy with risperidone in autistic children and adolescents.