The benefits of separating drug prescribing and dispensing are still unclear, in particular when drug consumption is characterized by important spillovers. We investigate the role of dispensing physicians in the consumption of antibiotics characterized by two opposite external effects: infection prevention and control, and bacterial resistance. We model the interaction between competing physicians (with and without dispensing of drugs) and patients exposed to bacterial infections and show that spatial effects of consumption may generate ambiguous results. Then, we propose an empirical exercise which exploits data from small geographic areas in Switzerland where two regimes – prescribing physicians and dispensing physicians – are possible.We consider spatial aspects of antibiotic consumption bymeans of combined spatial lag and spatial error econometric estimators for panel data (SARAR models). We find evidence that dispensing practices increase antibiotic use after controlling for determinants of demand and access, and spatial effects. Whether dispensing practices lead to an increase of antibiotic consumption beyond socially optimal levels is unclear and requires further research.