Prescribing quality assessment is increasingly used in improvement programs and pay-for-performance policies. Within the area of diabetes many quality indicators have been developed. Some measure prescribing on a general level, e.g. calculating percentages of patients prescribed any statins. Others are very specific, e.g. percentages of patients with an elevated LDL-cholesterol in whom lipid-lowering treatment is started unless contraindicated or return to control within 3 months. Although the latter seems more precise, we question how far one should go in developing such indicators. Using the example of diabetes treatment, we highlight the need, opportunities, and feasibility of assessing prescribing quality in the context of individualised treatment. We conclude that it is not realistic to develop indicators that take all possible aspects of therapy non-response, intolerance and patient preferences into account. We do recommend further development of indicators that better address subpopulations in need of adjusted treatment, such as elderly or patients with comorbidity.