Background To investigate the costs of treatment with basal insulin (insulin NPH [NPH], insulin glargine [IG], insulin determir [IG]), and premixed insulin (PM) in routine clinical care. Methods Cohort study based on data from the Swedish National Diabetes Register, including 5077 insulin-naïve men and women with type 2 diabetes, resident in a distinct geographical region of Sweden. Patients were included between 1 July 2006 and 31 December 2009 and followed for 12 months. All drug- and healthcare-related costs, stratified by diabetes-related or non-diabetes care contacts, were quantified and compared to baseline. Results Initiation of insulin treatment generally entails increased diabetes-related health care contacts and treatment costs, and decrease in health care costs. The median changes in costs were generally smaller than the mean changes, reflecting great variations between patients. The treatment costs were higher for IG, ID and PM compared with NPH, although higher age, history cardiovascular disease and diabetes complications as well as higher diabetes-related and other treatment costs were independent predictors. Overall, only PM (but not IG or ID) were associated with higher diabetes-related health care costs, although these were also independently predicted by cardiovascular morbidity and markers of complicated diabetes. Conclusions This study demonstrates that the initiation of insulin in patients with type 2 diabetes in clinical practice leads to increased health care contacts, overall and treatment costs, but also generally results in a decrease in health care costs. The diabetes-related treatment cost was lowest using NPH insulin but only premixed insulin was associated with higher diabetes-related health care costs than NPH. Electronic supplementary material The online version of this article (doi:10.1186/s40842-015-0017-1) contains supplementary material, which is available to authorized users.