Stromal cells have been studied extensively in the primary tumor microenvironment. In addition, mesenchymal stromal cells may participate in several steps of the metastatic cascade. Studying this interaction requires methods to distinguish and target stromal cells originating from the primary tumor versus their counterparts in the metastatic site. Here we illustrate a model of human tumor stromal cell-mouse cancer cell coimplantation. This model can be used to selectively deplete human stromal cells (using diphtheria toxin, DT) without affecting mouse cancer cells or host-derived stromal cells. Establishment of novel genetic models (e.g., transgenic expression of the DT receptor in specific cells) may eventually allow analogous models using syngeneic cells. Studying the role of stromal cells in metastasis using the model outlined above may take 8 weeks.