PURPOSE To determine the number of times patients have brand and generic products substituted under Australia's Pharmaceutical Benefits Scheme (PBS) brand substitution policy. METHODS A retrospective cohort study was conducted using Repatriation Pharmaceutical Benefits Scheme (RPBS) pharmacy claims data. Department of Veterans' Affairs (DVA) treatment card holders with at least two dispensings of atenolol, citalopram, enalapril, metformin, omeprazole or ramipril between 1 January 2001 and 28 February 2006 were included. Patients were followed from first dispensing until death, cessation or study end. The main outcome measure was the number of substitutions per patient during follow-up. Based on this, patients were defined as non-switchers, brand substitution or multiple switchers. RESULTS Data for 160,145 patients were analysed. Overall more than 80% of patients either had no switches or demonstrated brand substitution. For all study drugs, patients were more likely to be non-switchers than have a brand substitution (RR range 2.6-9.4, p < 0.0001) and were more likely to be non-switchers than multiple switchers (RR range 3.2-35.9, p < 0.0001). Patients who switched were more likely to have a brand substitution than multiple switches (RR range 1.2-3.8, p < 0.0001). Multivariate logistic regression showed greater odds of being a multiple switcher with increasing number of prescribers and dispensing pharmacies, and increasing length of follow-up. CONCLUSIONS Most patients in this study did not substitute products, and those who did were more likely to demonstrate brand substitution than have multiple switches. These results suggest that the brand substitution policy is having its intended effect for most patients.