Transcobalamin I and transcobalamin III have been purified approximately 6,000,000- and 3,000,000-fold, respectively, from normal human plasma using a purification scheme consisting of immunoadsorption, dialysis against 7.5 M guanidine HCl to remove endogenous vitamin B12, and affinity chromatography on vitamin B12-Sepharose. The two proteins were separated from each other subsequently by chromatography on DEAE-cellulose. The vitamin B12-binding protein present in granulocytes obtained from normal subjects has been purified approximately 5000-fold using affinity chromatography on vitamin B12-Sepharose as the sole purification technique. The final preparations of all three proteins were homogeneous based on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Transcobalamin I and transcobalamin III belong to the R-typed class of vitamin B12-binding proteins and are indistinguishable from each other, and from the human granulocyte, milk, and saliva R-type vitamin B12-binding proteins, when studied by immunodiffusion with rabbit anti-human milk vitamin B12-binding protein sera. The carbohydrate compositions, expressed as moles of carbohydrate per mole of vitamin B12, of transcobalamin I, transcobalamin III, and the normal granulocyte vitamin B12-binding protein, respectively, are: sialic acid, 18, 11, 11; fucose, 9, 20, 24; galactose, 41, 51, 46; mannose, 24, 22, 20; galactosamine, 2, 2, 2; and glucosamine, 46, 54, 46. The high sialic acid content of transcobalamin I appears to account for the fact that this protein elutes after transcobalamin III and the normal granulocyte vitamin B12-binding protein during chromatography on DEAE-cellulose. This observation provides support for the hypothesis that differences among the R-type vitamin B12-binding proteins are due to differences in carbohydrate content. The similarities in carbohydrate composition and other properties of transcobalamin III and the granulocyte vitamin B12-binding protein provide support for the hypothesis that human plasma transcobalamin III is derived from granulocytes. The differences observed between transcobalamin I and the normal granulocyte vitamin B12-binding protein suggest that transcobalamin I may not be derived from granulocytes.