Classically, most cell culture experiments have been performed under adherent 2D conditions. Cells in the human body grow within an organized 3D matrix, surrounded by other cells. The behavior of individual cells is controlled through their interactions with their immediate neighbors and the extracellular matrix. The complex summation of these multiple signals determines whether a given cell undergoes differentiation, apoptosis, proliferation, or invasion. In 2D culture many of these complex interactions are lost. As a result, there are a growing number of studies which report differences in phenotype, cellular signaling, cell migration, and drug responses when the same cells are grown under 2D or 3D culture conditions. One potential application of these techniques is to anticancer drug discovery, which has long been hampered by the lack of good preclinical models. Compounds with good antitumor activity in 2D cell culture models often fail to translate into the clinic. Here we suggest that the response of cancer cells to drugs is determined in part by the 3D tumor microenvironment and discuss models to re-create the 3D tumor microenvironment in vitro. It is likely that the adoption of these and other 3D models will allow us to more closely re-create the behavior of the tumor in vivo which may lead to identifying better anticancer drug candidates at an earlier stage of development.