The effect, if any, of 3 [human] lymphocyte separation techniques on the composition and functional characteristics of the purified cell suspensions was studied. Each separation technique yielded a cell population with highly specific and reproducible characteristics. The Ficoll-Hypaque technique led to good lymphocyte yields, but low yields of sheep erythrocyte-rosetting (E-rosetting) T [thymus-derived] cells; the separated cell population responded less to phytohemagglutinin [PHA]. The glass sand filtration technique led to lowest overall yield of small lymphocytes and of EAC[human erythrocytes, antibody, complement]-rosetting cells. There was significantly lower total yield of E-rosetting T cell; the separated lymphocyte suspension had excellent purity, had a relatively high percentage of E-rosetting T cells, and they responded well to PHA and pokeweed mitogen (PWM). The Technicon separation involving magnetic removal of phagocytic cells by exposure to Fe particles consistently led to large yields of small lymphocytes with good purity, the largest total harvests of E-rosetting T cells, and EAC-rosetting cells. The separated population had the highest percentage of E-rosetting cells and responded very well to PHA and PWM. Lymphocyte losses during purification are not nonspecific. The choice of the separation technique profoundly affects the characteristics of the purified lymphocyte population obtainable.