Clonogenic tumor cells contained within hematopoietic stem cell (HPC) grafts may contribute to relapse following autologous transplantation. Graft purging involves either in vivo or ex vivo HPC manipulation in order to reduce the level of tumor cell contamination. Some phase II trials suggest that patients who receive purged products may have a superior transplant outcome. Phase I trials demonstrate the feasibility of purging methods including ex vivo graft incubation with chemotherapeutic drugs, monoclonal antibodies and complement, and CD34+ cell selection. A phase II trial in follicular non-Hodgkin's lymphoma demonstrates that patients who receive HPC products purged negative for bcl-2 gene rearrangements have a superior outcome, compared with patients who receive polymerase chain reaction (PCR)-positive products. This finding, however, has not been confirmed in a randomized trial. HPC purging has demonstrated no benefit in a phase III trial in myeloma. Phase II trials in acute myelogenous leukemia show comparable outcomes for patients who receive either purged or unpurged HPC grafts. Limitations of purging include possible progenitor cell loss, delayed engraftment, and qualitative immune defects following transplant. Data to justify routine use of HPC graft purging are insufficient. Phase I and II data support development of phase III trials of both in vivo and in vitro purging methods.