Background and objectives: in Canada, the deferral for men who have sex with men (MSM) was decreased from a permanent deferral to a 5-year then a 12-month deferral. Current HIV testing can detect an HIV infection in donated blood within 2 weeks of exposure; thus, a 12-month deferral may be unnecessarily restrictive. We aimed to estimate the residual risk of HIV if the deferral were further decreased to 3 months. Materials and methods: using a deterministic model with stochastic Monte Carlo simulation, residual risk of HIV was the sum of testing error, assay sensitivity and window-period risks. Data inputs were estimated from donor surveillance, donor surveys and published data. Residual risk was modelled at baseline and using three scenarios: (1) most likely - non-compliance, HIV prevalence and incidence rates of MSM are unchanged; (2) optimistic - non-compliance improves by 50%; and (3) pessimistic - non-compliance, HIV prevalence and incidence rates of MSM all double. Results: HIV residual risk at baseline was 1 in 36·0 million donations (95% CI 1 in 1 504 907 million, 10·5 million); in the most likely scenario 1 in 34·2 million (1 in 225 534 million, 8·7 million); in the optimistic scenario 1 in 36·0 million (1 in 282 618 million, 9·5 million); in the pessimistic scenario 1 in 16·7 million (1 in 39 469 million, 6·0 million). All confidence intervals overlapped. Conclusion: with very low modelled risk under a 12-month deferral, the additional risk with a 3-month deferral is very low. This is true even with a pessimistic scenario.