OBJECTIVES: The contribution of weight change to the economic burden of T2DM is unclear. This study investigated associations between weight change and HCRU in patients with T2DM initiating new antidiabetic drug class.  METHODS: Patients with T2DM initiating new antidiabetic drug classes (first-line, switch or add-on) between 01/01/05-01/01/12 were identified in UK Clinical Practice Research Datalink primary care (PC) records linked with Hospital-Episode-Statistics. Baseline characteristics were assessed before drug class initiation and weight change (index date) was observed 6 months after. HCRU was followed up to one year after index and included diabetes-related PC contacts and prescriptions, and all-cause hospitalisation days. Weight change was categorized as: <3.0% change (weight-neutral),  3.0%-5.4% gain (gain+),  ≥5.5% gain (gain++), 3.0%-5.4% loss (loss+),  ≥5.5% loss (loss++). Comparisons between weight groups were conducted using negative binomial regression. RESULTS: Of 9031 patients with mean age 59.2 years and 57.4% men; 54.3% were weight-neutral, 12.0% gain+, 10.2% gain++, 12.8% loss+, 10.7% loss++. Mean baseline BMI ranged from 30.1 kg/m (gain++) to 33.8 kg/m (loss++). Follow-up crude mean PC contacts for all patients was 4.6 per person-year (PPY), mean prescriptions 15.3 PPY, mean hospitalisation days 1.5 PPY. For the weight-neutral group, crude mean PC contacts, prescriptions and hospitalisation days were 4.6, 14.9 and 1.2 PPY, respectively. In initial adjusted comparisons to weight-neutral patients, gain++ had more PC contacts, gain+ and gain++ had more prescriptions whereas loss+ patients had fewer. Gain+ and loss++ patients had more hospitalisation days than weight-neutral patients.  CONCLUSIONS: T2DM patients gaining weight after initiating new antidiabetic treatment may have increased health care resource use compared to weight-neutral patients. Higher weight loss, possibly due to underlying health problems, may also be associated with increased resource use. Causal interpretations of these results require detailed information on social and medical factors driving resource use in diabetic patients that were unavailable.