OBJECTIVES: To examine the effect of smoke-free prison policy implementation in November 2018 on purchasing patterns in the prison canteen (shop). DESIGN: Interrupted time series. SETTING: All 12 closed, publicly run prisons in Scotland, UK. PARTICIPANTS: People in custody (PiC) between August 2018 and end of March 2019 (n=11 944). INTERVENTIONS: Implementation of smoke-free prisons policy. OUTCOME MEASURES: Total spent on all products, nicotine-related products, and food and beverage products per week. METHODS: Canteen data were provided for the period July 2018-September 2019 by the Scottish Prison Service. In a series of generalised linear mixed effects models, the amount spent before and after implementation of smoke-free prison policy was compared for all purchases in the time period, and for PiC identified as 'smokers' and 'non-smokers' from their pre-implementation tobacco purchasing patterns. RESULTS: The amount spent on nicotine-related products significantly decreased from pre-implementation to post implementation (incident rate ratio (IRR) 0.40; 99% CI 0.33 to 0.51, p<0.001). However, total canteen spend did not change over the study period (IRR 0.92; 99% CI 0.84 to 1.00). Post implementation about 25% of previous 'smokers' total canteen spend was on nicotine-related products. The amount spent by previous 'smokers' on food and beverages increased from £8.67 (99% CI 8.23 to 9.13) pre-implementation to £10.24 in the post implementation period (99% CI 9.58 to 10.90). CONCLUSION: Although the amount of money previous 'smokers' in prison spent on nicotine-related products decreased after smoke-free policy, nicotine products still account for a large proportion of canteen spend among PiC in smoke-free prisons in Scotland. Results indicate that many PiC may continue to use nicotine in smoke-free prisons where e-cigarettes are permitted.