With the beginning of this millennium it has become fashionable to only follow 'evidence-based' practices. This generally-accepted approach cruelly negates experience or intelligent interpretation of pathophysiology. Another problem is that the great 'meta-analysts' of the present era only accept end points that they consider 'hard'. In the metabolic and nutritional field these end points are infection-related morbidity and mortality, and all other end points are considered 'surrogate'. The aim of this presentation is to prove that this claim greatly negates the contribution of more-fundamentally-oriented research, the fact that mortality has multifactorial causes, and that infection is a crude measure of immune function. The following problems should be considered: many populations undergoing intervention have low mortality, requiring studies with thousands of patients to demonstrate effects of intervention on mortality; nutrition is only in rare cases primary treatment, and in many populations is a prerequisite for survival rather than a therapeutic modality; once the effect of nutritional support is achieved, the extra benefit of modulation of the nutritional support regimen can only be modest; cost-benefit is not a valid end point, because the better it is done the more it will cost; morbidity and mortality are crude end points for the effect of nutritional intervention, and are influenced by many factors. In fact, it is a yes or no factor. In the literature the most important contributions include new insights into the pathogenesis of disease, the diminution of disease-related adverse events and/or functional improvement after therapy. In nutrition research the negligence of these end points has precluded the development and validation of functional end points, such as muscle, immune and cognitive functions. Disability, quality of life, morbidity and mortality are directly related to these functional variables. It is, therefore, of paramount importance to validate functional end points and to consider them as primary rather than surrogate end points.