Aims: Insulin is normally added to oral glucose-lowering drugs in people with type 2 diabetes when glycaemic control becomes suboptimal. We evaluated outcomes in people starting insulin therapy with neutral protamine Hagedorn (NPH), detemir, glargine or premixed insulins. Methods: Insulin-nïve people with type 2 diabetes (n = 8009), ≥ 35 years old, HbA 1c≥ 6.5% and begun on NPH (n = 1463), detemir (n = 357), glargine (n = 2197) or premix (n = 3992), were identified from a UK database of primary care records (The Health Improvement Network). Unadjusted and multivariate-adjusted analyses were conducted, with persistence of insulin therapy assessed by survival analysis. Results: In the study population (n = 4337), baseline HbA 1c was 9.5 ± 1.6%, falling to 8.4 ± 1.5% over 12 months (change -1.1 ± 1.8%, p < 0.001). Compared with NPH, people taking detemir, glargine and premix had an adjusted reduction in HbA 1c from baseline, of 0.00% (p = 0.99), 0.19% (p < 0.001) and 0.03% (p = 0.51). Body weight increased by 2.8 kg overall (p < 0.001), and by 2.3, 1.7, 1.9, and 3.3 kg on NPH, detemir, glargine and premix (p < 0.001 for all groups); insulin dose at 12 months was 0.70 (overall), 0.64, 0.61, 0.56 and 0.76 U/kg/day. After 36 months, 57% of people on NPH, 67% on glargine and 83% on premix remained on their initially prescribed insulin. Discussion and Conclusion: In routine clinical practice, people with type 2 diabetes commenced on NPH experienced a modest disadvantage in glycaemic control after 12 months compared with other insulins. When comparing the insulins, glargine achieved best HbA 1c reduction, while premix showed greatest weight gain and the highest dose requirement, but had the best persistence of therapy.