Percoll (colloidal silica coated with polyvinylpyrrolidone) and Ficoll (MW 400,000) were used to separate erythrocytes infected with Plasmodium yoelii and Plasmodium berghei from uninfected red blood cells. Samples of blood collected from mice in different phases of malarial infection were overlaid on cushions of 55% Percoll, 20% Ficoll, or 28% Ficoll, respectively, centrifuged, and the interphase layers compared. The best yield of parasitized erythrocytes (PE) was achieved using Percoll when about 95% of the erythrocytes infected by the late developmental forms of the parasites (late trophozoites, schizonts, and gametocytes) were recovered from the gradient interphase, irrespective of the phase of the infection and the number of young erythrocytes in the sample. No alteration of antigenicity (assessed by immunofluorescence) or of osmotic fragility (over the range of 160–460 mOsm) could be detected in PE separated by Percoll or by Ficoll. In addition, parasites separated on Percoll gradients showed no significant ultrastructural changes and retained their normal infectivity to mice. Although both gradient media could be used for the separation of Plasmodium-infected erythrocytes, Percoll presented some advantages over Ficoll. Apart from the better reproducibility of the separation of high yields of very pure PE obtained with Percoll, its lower viscosity allowed easier handling, and lower centrifugal forces were needed to enable the cells to reach their isopycnic positions. Thus, Percoll fulfilled many of the criteria for an ideal density gradient medium. Parasitized erythrocytes were isolated by an easy, reproducible, and inexpensive procedure, and separated cells retained their normal structure, antigenicity, and infectivity.