We tested two positive selection techniques for separation of CD34+ cells from bone marrow and analyzed the yields of CD34+ cells, BFU-E, CFU-GM, CFU-MK and LTC-IC after selection and expansion. An immunoadsorption procedure (CellPro) and an immunomagnetic (Baxter) CD34+ cell separation method were employed to purify the same bone marrow samples from seven normal subjects. Mean yields of CFU-GM and CFU-MK and absolute numbers of LTC-ICs were not different in the two purified cell populations. In contrast, the mean recovery of BFU-E was significantly lower for the immunoadsorption (21 +/- 14%) than for the immunomagnetic technique (44 +/- 27%). After separation, CD34+ cells were evaluated in 10-day liquid cultures for their expansion capacity in terms of total cells and progenitors. The expansion capacity of progenitors such as CFU-GM, CFU-MK and especially BFU-E selected by immunoadsorption was higher than the capacity of progenitors obtained by immunomagnetism, although final total and progenitor cell numbers are similar. Our results suggest that the populations separated by the two techniques differ mainly in the expansion capacity of progenitors and in the recovery of BFU-E after the selection procedure. These differences between two methods, which already are widely employed in research and in clinical transplantation, should be taken into account when considering the aims of the experiments.