Fc gamma RIII (CD16), a receptor for complexed IgG, is encoded by two very homologous genes: Fc gamma RIIIA and Fc gamma RIIIB. NK cells and macrophages express Fc gamma RIIIa, whereas only neutrophils constitutively express Fc gamma RIIIb. In a previous study we found that soluble (s)Fc gamma RIII in plasma seemed to originate only from neutrophils. However, CD16 mAb, directed against different epitopes of Fc gamma RIII, precipitated a glycoprotein from plasma of homozygous Fc gamma RIIIB gene-deficient donors. This glycoprotein migrated in a similar way as did released Fc gamma RIIIa derived from NK cells, whereas Fc gamma RIIIa released by cultured monocytes migrated differently and appeared to be more heavily glycosylated on SDS-PAGE. After deglycosylation, the M(r) of the plasma sFc gamma RIIIa was similar to that of released Fc gamma RIIIa. Moreover, V8-protease maps were identical. Therefore, we conclude that sFc gamma RIIIa is also present in plasma and is derived from NK cells. Because sFc gamma RIII levels are hardly detectable in the plasma of most homozygous Fc gamma RIIIB gene-deficient donors, we suspect that the sFc gamma RIIIa level is negligible compared with the level of sFc gamma RIIIb in plasma of healthy donors. Two patients with an NK cell lymphocytosis had a high plasma level of sFc gamma RIIIaNK. Furthermore, high levels of sFc gamma RIIIaNK were found in plasma of two patients with rheumatoid arthritis. Thus, the level of sFc gamma RIIIaNK might reflect either an increase in circulating NK cells or an enhanced release of Fc gamma RIIIaNK in certain diseases. This study shows that an assay that discriminates between sFc gamma RIIIa and sFc gamma RIIIb is necessary for the interpretation of sFc gamma RIII levels in patients.