AIMS: To examine the rates of diabetic kidney disease (DKD) progression and associated factors, we undertook a study of estimated glomerular filtration rate (eGFR) in a historical cohort of UK primary care patients with type 2 diabetes mellitus (T2DM) and associated DKD from the Clinical Practice Research Datalink. METHODS: Our eligible population were patients with definitive T2DM from a recorded diagnostic code with either a diagnosis of chronic kidney disease (CKD) or renal function test values and renal abnormalities consistent with a CKD diagnosis, identified between 1 October 2006 and 31 December 2011. Only patients with albuminuria results reported in mg/l were used for the longitudinal statistical analyses of the eGFR rate of change using multilevel models. RESULTS: We identified 111,030 patients with T2DM. Among them 58.6% (95% confidence interval (CI): 58.3-58.9) had CKD and 37.2% (95% CI: 36.9-37.5%) had presumed DKD at baseline. Only 19.4% of patients had urinary albumin test results expressed as mg/l in the year prior to index date. Almost two-thirds (63.8%) of patients with T2DM and presumed DKD received prescriptions for angiotensin-converting enzyme (ACE) inhibitors or angiotensin type 1 receptor blockers (ARB) or both. Time-dependent variables that predict subsequent eGFR decline include increased albuminuria, time from index date and older age. CONCLUSION: Only a minority of diabetic patients with DKD had quantitative albuminuria assessments. The relatively low proportion of DKD patients with ACEi or ARB prescriptions suggests a gap between healthcare practice and available scientific evidence during the study period. Increased albuminuria and older age were the most consistent predictors of subsequent eGFR decline.