The clinical efficacy of anti-angiogenic therapies has been difficult to predict, and biomarkers that can predict responsiveness are sorely needed in this era of personalized medicine. CVX-060 is an angiopoietin-2 (Ang2) targeting therapeutic, consisting of two peptides that bind Ang2 with high affinity and specificity, covalently fused to a scaffold antibody. In order to optimize the use of this compound in the clinic the construction of a predictive model is described, based on the efficacy of CVX-060 in 13 cell line and 2 patient-derived xenograft models. Pretreatment size tumors from each of the models were profiled for the levels of 27 protein markers of angiogenesis, SNP haplotype in 5 angiogenesis genes, and somatic mutation status for 11 genes implicated in tumor growth and/or vascularization. CVX-060 efficacy was determined as tumor growth inhibition (TGI%) at termination of each study. A predictive statistical model was constructed based on the correlation of these efficacy data with the marker profiles, and the model was subsequently tested by prospective analysis in 11 additional models. The results reveal a range of CVX-060 efficacy in xenograft models of diverse tissue types (0-64% TGI, median = 27%) and define a subset of 3 proteins (Ang1, EGF, Emmprin), the levels of which may be predictive of TGI by Ang2 blockade. The direction of the associations is such that better efficacy correlates with high levels of target and low levels of compensatory/antagonizing molecules. This effort has revealed a set of candidate predictive markers for CVX-060 efficacy that will be further evaluated in ongoing clinical trials.