We tested four negative and two positive selection methods for separation of CD34+ cells from mobilized blood cells, and analysed fold-enrichment, purity and recovery of CD34+ cells after selection procedures. The elimination of mature CD34- cells was achieved by adhesion to nylon-wool fibre (5.9 +/- 1.0 mean fold-enrichment and 65.2 +/- 2.3 mean recovery of CD34+ cells). Standard or modified Ficoll-Hypaque and Percoll density gradients, as well as phagocytosis with magnetic beads, were less effective in eliminating CD34- cells, both purity and fold-enrichment of CD34+ cells being lower than those obtained with separation by nylon-wool. Both positive selection methods tested. Ceprate and MiniMacs System, generated highly purified CD34+ cell populations ranging from 80% to 90%. The recovery of CD34+ cells was optimal with MiniMacs (77.9 +/- 3.6) and low with Ceprate (28.8 +/- 2.8). Based on these results, in two large-scale experiments we combined nylon-wool fibre and MiniMacs System in a two-step separation procedure obtaining a 36.9 +/- 2.6 mean fold-enrichment and a 50.5 +/- 0.3 mean recovery of CD34+ cells. In this way we achieved optimal enrichment and recovery of CD34+ cells, with a substantial saving of cost compared to either selection method alone.