In order to improve medication safety, more epidemiological data on the prevalence and clinical relevance of drug interactions are required. We developed an interface for mass analysis using the Clinical Decision Support Software (CDSS) MediQ and a multidimensional classification (Zurich Interaction System (ZHIAS)) incorporating the Operational Classification of Drug Interactions (ORCA). These were applied to 359,207 cross-sectional prescriptions from 84,607 psychiatric inpatients collected through the international AMSP program. MediQ issued 2,308 "high" and 71,112 "average" danger interaction alerts. Among these, after ORCA reclassification, there were 151 contraindicated and 4,099 provisionally contraindicated prescriptions. The ZHIAS provided further detailed categorical information on recommended management and specific increased risks (QTc prolongation being the most frequent one) associated with interactions. We developed a highly efficient solution for the identification and classification of drug interactions in large prescription data sets; this solution may help to reduce the frequency of overalerting and improve acceptance of the efficacy of CDSS in reducing the occurrence of potentially harmful drug interactions.