The IMPROVE (TM) study is a multinational, open-label, non-randomised, 26-week observational study assessing the safety and effectiveness of biphasic insulin aspart 30 (BIAsp 30) treatment in type 2 diabetes in routine clinical practice. The principal aims of this report were to characterise the baseline population and physicians' treatment decisions. Methods: Patients with type 2 diabetes who required insulin and whose physician had decided to initiate BIAsp 30 were eligible. At baseline, demographic data and detailed medical histories were collected and physicians recorded their reasons for starting BIAsp 30, the glycaemic targets set and the regimens chosen. Results: Data from 51,286 patients were included in analyses. Baseline glycaemic control was poor in all eight countries in the present analysis and in all prestudy treatment groups [no therapy, oral antidiabetic drugs (OADs) only, insulin with or without OADs], and the rates of vascular complications were high. Although the management of each of the three main measures of glycaemic control were key reasons for starting BIAsp 30, target-setting for postprandial glucose levels was variable. A twice-daily regimen was used to start BIAsp 30 therapy for 80% or more of patients. Conclusions: The IMPROVE (TM) baseline data reaffirm the global nature of poor glycaemic control in type 2 diabetes and echo the concerns that initiation of therapy, particularly insulin, is commonly delayed in clinical practice. Although postprandial glucose control was a key driver for physicians' choice of BIAsp 30, this was not consistently reflected in the targets set.