The efficiency of washing liquid-stored red blood cells [human] and red blood cells frozen with high or low glycerol concentrations was evaluated by measuring the recovery of red blood cells in vitro, supernatant Hb, extracellular K and red blood cell K levels, supernatant osmolality, residual 125I albumin, glycerol, hypoxanthine and di-2-ethylhexyl phthalate (DEHP) levels. Four commercial washing systems were studied, three which used NaCl solutions with serial or continuous-flow centrifugation and one which used sugar solutions and dilution/agglomeration. Washing was most efficient using NaCl solutions in the IBM Blood Processor, an automated serial centrifugation procedure and in the Fenwal Elutramatic, a continuous-flow centrifugation procedure. Less efficient washing was achieved in the Haemonetics Processor 15, a continuous-flow centrifugation procedure and the least efficient washing occurred using the original and modified dilution/agglomeration procedures. To achieve the most efficient washing, 3 principles must be utilized: concentration of the red blood cells to hematocrit values of 90%, prior to washing or freezing. Liquid-stored red blood cells concentrated to hematocrit values of 90 vol% should be diluted with hypertonic NaCl solutions prior to recovery and washing. Red blood cells containing 20% or 40% wt/vol glycerol should be diluted with hypertonic NaCl solutions before recovery and washing. Finally, on-line dilution should be achieved in the washing systems that use continuous-flow centrifugation.