Despite advances in understanding of the pathophysiology of type 2 diabetes and the abundance of expert reviews on its treatment, a large gap remains between what is generally believed and what is actually known about treatments for diabetes and their long-term effects. Over a relatively short recent interval the ADA and the EASD have jointly published successive consensus algorithms for the management of hyperglycaemia in patients with type 2 diabetes [1,2]. The authors of both the 2006 and 2009 consensus algorithms acknowledged in their introductions that while numerous reviews on the treatment of type 2 diabetes have been published in recent years, 'practitioners are often left without a clear pathway of therapy to follow' [1,2]. The most recent algorithm, with its two-tiered approach to metabolic management, has clearly not solved this problem and has itself provoked further debate and commentary [3,4], including most recently the critique by Schernthaner and colleagues [5] published in the current issue of Diabetologia.The most recent contributors, rather than proposing a further, newer algorithm for diabetes treatment (perhaps we should be grateful that further guidelines have not been proposed), take issue with the 'expert opinion' approach adopted by the ADA/EASD authors and argue instead for an evidence-based strategy, taking into account the impact of various treatments on clinically important endpoints (such as macrovascular events). They put forward the case for individual optimisation of care of the patient with the goals of sustained blood glucose control and the reduction of complications and cardiovascular risk. Clearly no sensible clinician would dispute this case, and this includes those who have generated the previous guidelines and algorithms. Following the ancient principle of 'first, do no harm', the overall goal for a good doctor must be the improved care and quality of life of the individual patient. However, equally clearly there is a problem with treatment algorithms for type 2 diabetes based on 'expert opinion'. The article by Schernthaner and colleagues [5] raises some important limitations inherent in this approach and, in the process, points to some of the broader challenges in modern diabetes care. Their conclusion is that the 2009 treatment algorithm is not in fact based on evidence. This debate is useful for those who care for patients with diabetes, in that it emphasises the importance of evidence, while drawing our attention to the wide areas where the answers are not yet known, and where the evidence does not yet exist and may not exist for some time. The prevalence of diabetes continues to increase and actual treatment decisions and choices face individual patients and their doctors on a daily basis. How these decisions are best made in the real world may ultimately be through a pragmatic individual patient plan, taking account of the best available evidence, the stage in life of the patient, the biological stage of progression of their diabetes and the relative priorities of...