Tumor cells frequently contaminate autologous stem cell products in patients having a variety of malignancies. Mobilized peripheral blood stem cells may be less contaminated with tumor cells than bone marrow harvests are, but they are still frequently infiltrated. Gene-marking studies using retroviral vectors provide evidence that tumor cells contained in autografts contribute to relapse in myeloid leukemia and neuroblastoma patients. Also clinical studies have shown that tumor cell contamination of autografts is associated with shortened disease-free survival; on the other hand, successful ex vivo purging of tumor cells is associated with superior clinical outcome. However, the presence of tumor cells in autografts or insufficient purging may correlate with the extent of systemic residual disease and/or tumor chemosensitivity; therefore, there is no direct evidence that reinfused tumor cells alone cause relapse. Particularly in patients having highly chemosensitive disease and no detected systemic residual disease following high-dose transplant chemotherapy, the relative number of tumor cells contained in autografts and eventually reinfused, may become a determining factor for clinical outcome. There are no randomized trials showing improved (disease-free) survival with purging. In the absence of such trials, the contribution of tumor cells in the stem cell autografts to subsequent relapse remains controversial.