Objective To assess the fidelity of general practitioners' (GPS) adherence to a long-term pragmatic trial protocol. Design Retrospective analyses of electronic primary care records of participants in the pragmatic cluster-randomised ADDITION (Anglo-Danish-Dutch Study of Intensive Treatment In People with Screen Detected Diabetes in Primary Care)-Cambridge trial, comparing intensive multifactorial treatment (IT) versus routine care (RC). Data were collected from the date of diagnosis until December 2010. Setting Primary care surgeries in the East of England. Study sample/participants A subsample (n=189, RC arm: n=99, IT arm: n=90) of patients from the ADDITION-Cambridge cohort (867 patients), consisting of patients 40-69 years old with screen-detected diabetes mellitus. Interventions In the RC arm treatment was delivered according to concurrent treatment guidelines. Surgeries in the IT arm received funding for additional contacts between GPS/nurses and patients, and GPS were advised to follow more intensive treatment algorithms for the management of glucose, lipids and blood pressure and aspirin therapy than in the RC arm. Outcome measures The number of annual contacts between patients and GPS/nurses, the proportion of patients receiving prescriptions for cardiometabolic medication in years 1-5 after diabetes diagnosis and the adherence to prescription algorithms. Results The difference in the number of annual GP contacts (β=0.65) and nurse contacts (β=0.15) between the study arms was small and insignificant. Patients in the IT arm were more likely to receive glucose-lowering (OR=3.27), ACE-inhibiting (OR=2.03) and lipid-lowering drugs (OR=2.42, all p values <0.01) than patients in the RC arm. The prescription adherence varied between medication classes, but improved in both trial arms over the 5-year follow-up. Conclusions The adherence of GPS to different aspects of the trial protocol was mixed. Background changes in healthcare policy need to be considered as they have the potential to dilute differences in treatment intensity and hence incremental effects.