Recent research has employed different analytical techniques to estimate the impact of the various long-term complications of type 2 diabetes on health-related utility and health status. However, limited patient numbers or lack of variety of patient experience has limited their power to discriminate between separate complications and grades of severity. In this study alternative statistical model forms were compared to investigate the influence of various factors on self-assessed health status and calculated utility scores, including the presence and severity of complications, and type of diabetes therapy. Responses to the EuroQol EQ-5D questionnaire from 4641 patients with type 2 diabetes in 5 European countries were analysed. Simple multiple regression analysis was used to model both visual analogue scale (VAS) scores and time trade-off index scores (TTO). Also, two complex models were developed for TTO analysis using a structure suggested by the EuroQol calculation algorithm. Both VAS and TTO models achieved greater explanatory power than in earlier studies. Relative weightings for individual complications differed between VAS and TTO scales, reflecting the strong influence of loss of mobility and severe pain in the EuroQol algorithm. Insulin-based therapy was uniformly associated with a detrimental effect equivalent to an additional moderate complication. Evidence was found that TTO values are not responsive in cases where 3 or more multiple complications are present, and therefore may underestimate utility loss for patients most adversely affected by complex chronic diseases like diabetes.