Risk assessment for nutrients assumes a single population with a normal distribution of indexes of requirements and excess. Toxic levels are by definition intakes above the upper level; for copper, however, because we lack noninvasive, sensitive biomarkers of storage or early damage from excess, excess is based on the infrequent occurrence of clinical disease, such as unexplained liver cirrhosis. We examine the limitations of this approach for copper given the very low prevalence of clinical and subclinical disease and suggest that the population risk for copper excess be based on hepatic copper loading as a potentially quantifiable measurement. The challenge ahead is to develop biomarkers that predict the population risk of elevated hepatic copper stores and thus the possibility of disease in a population.