Rabbit transferrin in vitro is shown to load ferrous iron at random on its specific binding sites. The release of iron to reticulocytes is shown to be an all-or-none phenomenon. The two monoferric transferrins have similar in vivo plasma iron clearance rates and tissue distribution. Diferric transferrin, while giving a similar tissue distribution of radioiron, has a plasma iron clearance rate approximately twice that of the monoferric transferrins at low plasma iron concentrations. This difference diminishes as the plasma iron concentration increases. These results are consistent with a progressively greater in vivo conversion of mono- to diferric transferrin as transferrin saturation increases. The in vivo plasma iron turnover in the rabbit increases progressively as the plasma iron increases, from a mean value of -0.8 mg/dl whole blood per d at a plasma iron concentration of 50 ug/dl to 2.0 at a plasma iron concentration of 300. The molecular behavior of transferrin and its iron over this range was investigated using '251-transferrin, [55Fe]monoferric transferrin, and [59Fe]diferric transferrin. The equilibrium distribution of transferrin between its apo-, mono-, and diferric moieties was similar to that predicted on the basis of the percent saturation and random distribution. Rate constants of iron loading and unloading calculated from the percent saturation and from the clearance rates of [55Fe]- monoferric and [59Fe]diferric transferrin were similar to those derived from changes in injected '251-apotransferrin. On the basis of these data, it is concluded that the plasma transferrin pool is nonhomogeneous and that the relative size of the mono- and diferric cycles depends on transferrin saturation. A formula is proposed for correcting the plasma iron turnover, thereby eliminating the effect of plasma iron concentration, so as to reflect directly the number of tissue transferrin receptors.