Hemolytic anemia in sickle disease involves both intravascular and extravascular destruction of erythrocytes. Since the latter presumably involves the reticuloendothelial system, we have examined interactions between sickle erythrocytes and macrophages. In erythrophagocytosis assays, 18.9 ± 7.2% of human marrow macrophages ingest sickle RBCs, while only 3.1 ± 2.1% ingest normal RBCs. This abnormality is not explained by reticulocytosis, and it is strongly dependent upon RBC density. The interaction between sickle RBCs and macrophages appears to be partly immunologic, since it is partially blocked by Fc receptor blockade. Also, admixture of sickle RBCs (pre-treated with rabbit anti-human-Ig) and Fc-receptor-bearing K562 cells results in 15.6 ± 10.6% K562-RBC rosette formation compared with only 0.5 ± 1.2% for normal RBCs. Regarding other factors that might promote erythrophagocytosis, sickle RBCs are found to spontaneously generate twice-normal amounts of dialdehyde byproducts of lipid peroxidation ('malondialdehyde' or MDA). Peroxide or reagent-MDA treatment of normal RBCs significantly enhances their phagocytosis, and MDA is at least 50 times more potent than other aldehydes studied here. Oxidative and immunologic effects may be related, since exposure of MDA-treated RBCs to immunoglobulin-containing human sera results in a further significant enhancement of erythrophagocytosis. For comparison of different sickle patients, an adherence assay demonstrates that sickle RBCs are 1.03 to 6.85 times more adherent to macrophages than are normal RBCs, and degree of adherence correlates significantly with irreversibly sickled cells (ISC) counts and hematologic variables reflecting hemolytic rate. We conclude that propensity for RBC interaction with macrophages is likely to be a determinant of hemolytic rate in sickle disease. Pertinent mechanisms appear to involve modification of RBC membranes by dialdehyde byproducts of excessive autoxidation and the abnormal acquisition of surface immunoglobulin on sickle RBCs, although participation of other membrane defects has not been excluded. Interestingly, the data further suggest the possibility that appearance of the 'senescence antigen' in old normal RBCs represents modification of the membrane by 'MDA'.