Technical Abstract: Sudden losses of managed honey bee (Apis mellifera L.) colonies are considered an important problem worldwide but the underlying cause or causes of these losses are currently unknown. In the United States, this syndrome was termed Colony Collapse Disorder (CCD), since the defining trait was a rapid loss of adult worker bees. We conducted a classification and regression tree (CART) analysis in order to understand the relative importance and inter-relations among different explanatory risk variables in explaining CCD. Using a CART model with a specified cost of misclassification, we obtained a tree with a sensitivity of 100% and a specificity of 67%. With this model, CCD was characterized mostly by a low level of coumaphos (a miticide commonly used by beekeepers) in brood and beebread, smaller colony size as measured but frames of bees, physiological disturbances which suggest decreased colony fitness (lower mass of the head and increased fluctuating asymmetry), and increased viral loads. The CART analysis provides further evidence that CCD is likely the result of several factors acting in concert, with the combination decreasing colony fitness and making afflicted colonies more susceptible to disease. This analysis highlighted several areas that warrant further attention, including the effect of sub-lethal pesticide exposure has on pathogen expression, the role of variability in bee tolerance to pesticides on colony survivorship, and the consequences of pathology on bee physiology.