NK cells are essential for health, yet little is known about human NK turnover in vivo. In both young and elderly women, all NK subsets proliferated and died more rapidly than T cells. CD56(bright) NK cells proliferated rapidly but died relatively slowly, suggesting that proliferating CD56(bright) cells differentiate into CD56(dim) NK cells in vivo. The relationship between CD56(dim) and CD56(bright) proliferating cells indicates that proliferating CD56(dim) cells both self-renew and are derived from proliferating CD56(bright) NK cells. Our data suggest that some dying CD56(dim) cells become CD16(+)CD56(-) NK cells and that CD16(-)CD56(low) NK cells respond rapidly to cellular and cytokine stimulation. We propose a model in which all NK cell subsets are in dynamic flux. About half of CD56(dim) NK cells expressed CD57, which was weakly associated with low proliferation. Surprisingly, CD57 expression was associated with higher proliferation rates in both CD8(+) and CD8(-) T cells. Therefore, CD57 is not a reliable marker of senescent, nonproliferative T cells in vivo. NKG2A expression declined with age on both NK cells and T cells. Killer cell Ig-like receptor expression increased with age on T cells but not on NK cells. Although the percentage of CD56(bright) NK cells declined with age and the percentage of CD56(dim) NK cells increased with age, there were no significant age-related proliferation or apoptosis differences for these two populations or for total NK cells. In vivo human NK cell turnover is rapid in both young and elderly adults.