Background: As in other societies, pharmaceutical expenditures in the Netherlands are rising every year. As a consequence, needs for cost control are often expressed. One possible solution for cost control could come through increasing generic substitution by pharmacists. We aim to analyse the extent and nature of substitution in recent years and estimate the likelihood of generic or branded substitution in Dutch pharmacies in relation to various characteristics. Methods: We utilized a linked prescription dataset originating from a general practitioner (GP) and a pharmacy database, both from the northern Netherlands. We selected specific drugs of interest, containing about 55,000 prescriptions from 15 different classes. We used a crossed generalized linear mixed model to estimate the effects that certain patient and pharmacy characteristics as well as timing have on the likelihood that a prescription will eventually be substituted by the pharmacist. Results: Generic substitution occurred at 25% of the branded prescriptions. Generic substitution was more likely to occur earlier in time after patent expiry and to patients that were older and more experienced in their drug use. Individually owned pharmacies had a lower probability of generic substitution compared to chain pharmacies. Oppositely, branded substitution occurred in 10% of generic prescriptions and was positively related to the patients' experience in branded use. Individually owned pharmacies were more likely to substitute a generic drug to a branded compared to other pharmacies. Antidepressant and PPI prescriptions were less prone to generic and more prone to branded substitution. Conclusion: Analysis of prescription substitution by the pharmacist revealed strong relations between substitution and patient experience on drug use, pharmacy status and timing. These findings can be utilised to design further strategies to enhance generic substitution.