A chronic problem in cancer research is the discrepancy between preclinical outcomes of new cancer therapies in mice compared to later clinical trial results. The trend is highly encouraging preclinical findings which completely fail in the clinic, at least at the randomized phase III level. Two ongoing strategies for improving preclinical models are using genetically engineered mouse models or patient-derived xenografts. We have adopted a different strategy: using imageable transplanted tumor cell lines for the development of orthotopic postsurgical models of systemic metastatic disease, both early and late stage in nature. The rationale is to try and replicate the more challenging circumstance of treating patients with metastatic disease rather than the historical approach of just treating primary tumors. Using a number of different human tumor models involving both early and advanced metastatic disease, several examples of results obtained using them will be summarized. They are: i) the finding that advanced macroscopic metastases are resistant or less sensitive to antiangiogenic drugs, in contrast to primary tumors and that the basis for this is a lack of angiogenesis in the metastases, where instead, they can exploit what is known as "vessel co-option" ie. tumors in vascular rich organs such as the lungs, liver, or brain co-opting their abundant existing vasculature; ii) the discovery of 'late' spontaneous brain metastases in mice with systemic disease whose survival is extended by therapy; iii) the potent efficacy of low dose metronomic chemotherapy regimens alone or in combination with an antiangiogenic drug to treat both early and late stage metastatic disease; and iv) the failure of postsurgical adjuvant antiangiogenic therapies of early stage disease using VEGF pathway inhibitors. Recent initiatives to develop similar models of metastatic disease, but using syngeneic mouse tumors for immuno-oncology drug studies will also be discussed.