OBJECTIVES: No representative data on the quality of drug prescription exist in Switzerland. We analysed potential drug interactions (PDI) in primary care based on prescription data from 2010 from three large health insurers. METHODS: In our study population we identified based on the national drug formulary 494 drug-drug interactions (DDI) classes. Of those, 41 were from drug interaction severity classes I (contraindicated) and II (potentially contraindicated) and classified as PDI. PDI were calculated for two indicators with different denominators, 1) the total number of DDI and 2) the total number of patients potentially exposed to a PDI. For each physician we calculated the probability that the number of caused PDI was unlikely (p-value between 0.05 and 0.01; ‘problematic prescription behaviour’) and very unlikely (p-value <0.01; ‘likely problematic prescription behaviour’) to be explained by chance. RESULTS: Of 3.13 million individuals 1.34% were exposed to at least one PDI; figures increased to 3.78% and 4.40% in females and males aged ≥70 years. Of 20,710 physicians 42% caused at least one PDI. With DDI being the denominator 6.2% und 3.4% of general practitioners and 0.9% und 0.5% of specialists were classified with a ‘problematic’ and ‘likely problematic prescription behaviour’. With the patient population being the denominator between 0.96% and 6.22% of physicians from all specialty groups had caused a PDI that was classified as ‘likely problematic’ (p-value <0.01). When combining both indicators 457 of 20‘720 physicians (2.2%) had a prescription behaviour causing a PDI that in comparison to peers was highly unlikely to be explained by chance (p-value <0.01). CONCLUSIONS: 2.2% of physicians in primary care prescribe during one year at least one drug combination that is classified as PDI. A linkage to other data sources would be needed to further classify PDI that put patients at serious risk for adverse drug reactions.