Abstract Aims To compare long‐term complications for people with type 2 diabetes mellitus (T2DM) following second‐line treatment in routine practice with sulphonylureas (SU), dipeptidyl peptidase‐4 inhibitors (DPP4i), or sodium‐glucose co‐transporter‐2 inhibitors (SGLT2i) added to metformin. Materials and Methods We used the RAPIDS microsimulation model to predict diabetes complications over 5 years after second‐line treatment initiation. We combined information on ‘real‐world’ treatment duration in England from the Clinical Practice Research Datalink with evidence on treatment effectiveness from Randomised Controlled Trials (RCTs). We estimated between‐treatment differences in the probabilities of end‐stage kidney disease (ESKD), heart failure hospitalisation (HF), diabetic eye disease, myocardial infarction (MI), and lower‐extremity amputation (LEA). Results The predicted probabilities of complications within 5 years were lower following second‐line treatment with SGLT2i compared to SU and DPP4i. The mean (95% CI) difference (reduction) in the predicted probability of ESKD following SGLT2i versus SU was −0.81% (−0.89, −0.73), and for SGLT2i versus DPP4i the corresponding difference was −0.87% (−0.95, −0.79). The reduction in the probability of HF following SGLT2i versus SU was −0.90% (−1.01, −0.80), and for SGLT2i versus DPP4i it was −0.95% (−1.06, −0.84). The corresponding differences in the probabilities of diabetic eye disease following SGLT2i versus SU were −1.41% (−1.57, −1.26), and for SGLT2i versus DPP4i was −0.44% (−0.59, −0.29). The predicted probabilities of LEA were similar across treatments. Pre‐existing CVD did not modify the predicted probabilities of complications. Conclusions For a general T2DM population, second‐line treatment with SGLT2i rather than SU or DPP4i can reduce the probability of complications within 5 years.