Seven methods of erythrocyte entrapment--six by hypotonic exchange loading and one by chlorpromazine-induced endocytosis--were evaluted for (1) efficiency of incorporation of beta-glucuronidase, inulin, and glucose; (2) presence of beta-glucuronidase on the erythrocyte surface, as detected by hemagglutination or complement-dependent lysis in the prsence of antibody to beta-glucuronidase; (3) in vitro leakage of entrapped markers; and (4) morphologic alterations by scanning electron microscopy. Dependent on the method of entrapment, the incorporation of markers ranged from 0.7% to 6.2% for beta-glucuronidase, 3.4% to 31% for inulin, and 1.2% to 12.2% for glucose. Maximal incorporation by hypotonic exchange loading occurred when the initial concentration of sodium chloride (150 mM) was reduced to 50 or 75 mM. However, beta-glucuronidase was detected on the erythrocyte surface by hemagglutination for these methods as well as a dialysis method of loading. Essentially no leakage of entrapped enzyme was detected (< 1%) for all methods, although up to 11% of entrapped glucose was released during a 3 hr incubation at 37 degrees C in buffered whole blood. Finally, entrapment methods requiring the greatest reduction in salt concentration resulted in the formation of echinocytes, whereas stomatocytes were observed after entrapment by methods requiring lesser salt dilutions. These results demonstrate the efficiency of entrapment and relative integrity of erythrocytes following various loadng procedures and suggest that in vitro assessment may provide a useful predictor of the imunogenicity and in vivo fate of erythrocyte-entrapped enzymes or other therapeutic agents.