OBJECTIVES: To understand how insulin is initiated, describe patients with type 2 diabetes (T2DM) initiating insulin in UK primary care, and look at changes in therapy, clinical outcomes and contact with health care professionals (HCPs) after insulin initiation.  METHODS: Patients aged ≥18 years with a code for T2DM initiating insulin (baseline) from May 2004 to May 2006 were identified in the THIN database. Patient characteristics, clinical measures, treatment and contact with HCPs were collected at baseline and during 6 months follow-up.   RESULTS: A total of 4045 patients met the inclusion criteria, 56% were male.  Patient characteristics at baseline [mean (SD)] were; age: 62.6(13.3) years, BMI: 30.2(6.5) kg/m2 and HbA1c: 9.6%(2.0%). The most prevalent co-morbidities were hypertension (63%), coronary heart disease (26%), hyperlipaedemia (24%), and depression (22%).  97.4% of patients had contact with their GP-surgery in the 6 months before insulin initiation, mean 7.1 contacts; 78.1% were referred to secondary care; and 18% were hospitalised. In the 12 months prior to insulin initiation 8.5% of patients had no prescriptions for oral antidiabetic medications (OAD), 13.4% for a single OAD class and 78.1% for ≥2 different OAD classes. 52.4% of patients initiated intermediate/long acting insulin therapy only, a further 41.6% pre-mix insulin only and the remainder initiated basal bolus (4.0%) or fast acting only (2.1%). During follow-up, 14.7% discontinued insulin therapy, 6.8% switched to a different regimen and 4.7% intensified therapy; adding mealtime insulin. Of the patients with a measurement during follow-up (n=3024), 17.3% achieved target HbA1c (<7%). Clinical outcomes during follow-up [mean (SD)] were; HbA1c: 8.3 (1.6)% and BMI: 30.7 (6.6) kg/m2. There was a slight increase in contact with HCPs during follow-up, mean 8.5 GP-surgery contacts.  CONCLUSIONS: Patients had elevated HbA1c at insulin initiation, simple regimens were favoured; during a 6 month follow-up a high proportion of patients failed to achieve glycaemic target.