Cancer is a heterogeneous disease consisting of different cellular communities which engage and adapt with the microenvironment to survive and proliferate. Cancer cell lines have proven successful drug discovery tools in the past, yet they have recognized limitations, specifically the lack of clinical predictive power owing to their restricted capacity to recapitulate cancer's diversity. In order to capitalize on the rise of precision medicine, we must fully consider tumor heterogeneity at the earliest stages of oncological drug development. This chapter outlines current theories explaining the origin and proposed genomic causes of heterogeneity before discussing the potential influences of nongenetic context-dependent mechanisms. We conclude by discussing the use of the patient-derived tumor xenografts as new preclinical tools that closely resemble the originating cancer sample in histological, molecular, and epigenetic spaces including their intratumor genomic architecture. These models capable of mimicking cancer's heterogeneity will have substantial utility in predicting clinical efficacy.