Allogeneic natural killer (NK) cells are used for adoptive immunotherapy after stem cell transplantation. In order to overcome technical limitations in NK cell purification and activation, the following study investigates the impact of different variables on NK cell recovery, cytotoxicity, and T-cell depletion during good manufacturing practice (GMP)-grade NK cell selection. Forty NK cell products were derived from 54 unstimulated donor leukaphereses using immunomagnetic CD3 T-cell depletion, followed by a CD56 cell enrichment step. For T-cell depletion, either the depletion 2.1 program in single or double procedure (D2.11depl, n = 18; D2.12depl, n = 13) or the faster depletion 3.1 (D3.1, n = 9) was used on the CliniMACS instrument. Seventeen purified NK cell products were activated in vitro by IL-2 for 12 days. The whole process resulted in a median number of 7.59 × 10(8) CD56(+)CD3(-) cells with both purity and viability of 94%, respectively. The T-cell depletion was significantly better using D2.11depl/2depl compared to D3.1 (log 4.6/log 4.9 vs. log 3.7; p < 0.01) and double procedure in two stages led always to residual T cells below 0.1%. In contrast D3.1 was superior to D2.11depl/2depl with regard to recovery of CD56(+)CD3(-) NK cells (68% vs. 41%/38%). Concomitant monocytes and especially IL-2 activation led to increased NK cell activity against malignant target cells compared to unstimulated NK cells, which correlated with both up-regulation of natural cytotoxicity receptors and intracellular signaling. Overall, wide variations in the NK cell expansion rate and the distribution of NK cell subpopulations were found. In conclusion, our results indicate that GMP-grade purification of NK cells might be improved by a sequential processing of T-cell depletion program D2.1 and D3.1. In addition NK cell expansion protocols need to be further optimized.