OBJECTIVES: Diabetes Mellitus is one of the most expensive common diseases. Because of the great socio-economic importance of this indication, it seems necessary to consider the expectations and needs of all patients with regard to treatment characteristics. To evaluate patient preferences stated-preference methods are increasingly used. The aim was to analyze and compare the available evidence from patient preference studies regarding the treatment characteristics and to display which target criteria are most important from the patients view. METHODS: A literature review in PubMed was conducted to identify stated preference studies in the indication of diabetes and show which properties of a treatment are relevant to the patient's benefit. By means of level difference calculation and a subsequent normalization uniform weights for the attributes of the studies were determined. Based on a final ranking the key criteria according to their value proposition from the perspective of affected populations are demonstrated. RESULTS: N=13 studies could be included in the analysis. By calculating the level difference and performing a normalization it was possible to obtain a uniform representation of all attributes of each study. The analysis includes the presentation and analysis of DCE-studies which have illustrated the determined coefficients. The studies show that blood sugar control, side effects (such as hypoglycemia, nausea/stomach discomfort, weight changes), long-term complications (especially cardiovascular disease) and mode of administration represent the most patient-relevant outcomes in diabetes therapy. Blood glucose control achieved the highest ranking within the relative importance score. CONCLUSIONS: Based on the findings of this systematic review the evidence of patient preferences in diabetes treatment is synthesized. The results show the relative importance based on a derived ranking score of the used patient-relevant endpoints. The determined ranking allows the comparison of the results of the identified preference studies and might solve the problem of scale heterogeneity.