Abstract Aims/hypothesis Type 2 diabetes mellitus can manifest over a broad clinical range, although there is no clear consensus on the categorisation of disease complexity. We assessed the effects of canagliflozin, compared with placebo, on cardiovascular and kidney outcomes in the CANagliflozin cardioVascular Assessment Study (CANVAS) Program over a range of type 2 diabetes mellitus complexity, defined separately by baseline intensity of treatment, duration of diabetes and glycaemic control. Methods We performed a post hoc analysis of the effects of canagliflozin on major adverse cardiovascular events (MACE) according to baseline glucose-lowering treatments (0 or 1, 2 or 3+ non-insulin glucose-lowering treatments, or insulin-based treatment), duration of diabetes (<10, 10 to 16, >16 years) and HbA_1c (≤53.0 mmol/mol [<7.0%], >53.0 to 58.5 mmol/mol [>7.0% to 7.5%], >58.5 to 63.9 mmol/mol [>7.5 to 8.0%], >63.9 to 69.4 mmol/mol [8.0% to 8.5%], >69.4 to 74.9 mmol/mol [>8.5 to 9.0%] or >74.9 mmol/mol [>9.0%]). We analysed additional secondary endpoints for cardiovascular and kidney outcomes, including a combined kidney outcome of sustained 40% decline in eGFR, end-stage kidney disease or death due to kidney disease. We used Cox regression analyses and compared the constancy of HRs across subgroups by fitting an interaction term (p value for significance <0.05). Results At study initiation, 5095 (50%) CANVAS Program participants were treated with insulin, 2100 (21%) had an HbA_1c > 74.9 mmol/mol (9.0%) and the median duration of diabetes was 12.6 years (interquartile interval 8.0–18 years). Canagliflozin reduced MACE (HR 0.86 [95% CI 0.75, 0.97]) with no evidence that the benefit differed between subgroups defined by the number of glucose-lowering treatments, the duration of diabetes or baseline HbA_1c (all p-heterogeneity >0.17). Canagliflozin reduced MACE in participants receiving insulin with no evidence that the benefit differed from other participants in the trial (HR 0.85 [95% CI 0.72, 1.00]). Similar results were observed for other cardiovascular outcomes and for the combined kidney outcome (HR for combined kidney outcome 0.60 [95% CI 0.47, 0.77]), with all p-heterogeneity >0.37. Conclusions/interpretation In people with type 2 diabetes mellitus at high cardiovascular risk, there was no evidence that cardiovascular and renal protection with canagliflozin differed across subgroups defined by baseline treatment intensity, duration of diabetes or HbA_1c. Graphical abstract