Immunomagnetic separation with anti-CD34 monoclonal antibodies and paramagnetic microbeads has been used to enrich hematopoietic stem cells from human bone marrow (BM) or mobilized peripheral blood mononuclear cells (PBMNC). The introduction of this technique also constitutes a new principle of tumor cell purging. The efficiency in terms of purging tumor cells from PBMNC was evaluated in seven different experiments. Mobilized (chemotherapy and G-CSF) PBMNC were collected from patients with solid tumors (n = 6) and multiple myeloma (n = 1) by leukapheresis using an automated MNC separation system and contaminated with 1% (n = 5) or 10% (n = 2) tumor cells from different epithelial cell lines being CD34-negative. The cell mixture was sensitized with anti-CD34 (9C5) antibodies and sheep anti-mouse IgG1 paramagnetic microspheres and enriched for CD34+ cells using an Isolex 50 magnetic separator. Purify of CD34+ cells was studied by flow cytometry (FACScan) and tumor cell depletion was evaluated by comparative human tumor cloning assays (HTCA) containing methylcellulose and agar. We achieved a median purity of CD34+ cells of 85.9% (range 69.8-92.9%) and a median yield of 48.1% (range 21.0-85.2%). From these data in each case the estimated log depletion of tumor cells was calculated and compared with the experimentally achieved (HTCA) log depletion (log delta depletion = log experimental depletion--log calculated depletion). In our experiments we achieved a median depletion of 2.75 log (range 1.55-3.69 log). When corrected for CD34+ cell yield of each experiment we observed a median 'yield corrected depletion' of 2.38 log (range 1.48-3.15 log). The following delta depletion values were obtained: +0.32 log (HTB 129, breast), +0.21 log (HTB 26, breast), +0.04 log (HTB 26) for experiments with higher experimental depletion, and -0.23 log (HTB 26), -0.9 log (HTB 26, PBMNC from patient with multiple myeloma), -0.82 log (HTB 131, breast) and -1.66 log (HTB 131) for lower depletion efficacy than calculated. These data suggest that depletion may depend on specific cell surface characteristics of tumor cells. Moreover, plasma factors (eg paraprotein) may also have some impact. In summary, the Isolex 50 provides a high purity of CD34+ cells and depletion of tumor cells was efficient. However, calculated and experimental purging efficiencies are not necessarily identical.