Several potentially important parameters influencing the disproportionate distribution of red cell flux and blood flow at a bifurcation are examined. These include parent vessel hematocrit, vessel diameter, suspending medium, cell distensibility, parent vessel blood flow rate, and local geometry. Measurements were performed on 20 to 100 micrometers bifurcations, fabricated such that all vessels of a given bifurcations have the same diameter. Suspensions of human red cells, hardened red cells, and mixtures of each in albumin-saline, Dextran 75, or plasma were flowed through the bifurcations and determinations of flow rates and discharge hematocrits were made for each of the channels. For the 20-micrometers channels, hematocrits were found using videophotometric techniques, and for the larger channels, hematocrits were measured directly from the exit streams. Flow rates for both were measured by meniscus travel downstream in small-diameter glass tubes. Within the limits of the present experiments, three of the variables proved to be of major importance: feed hematocrit, tube diameter, and flow-rate distribution. It was clearly demonstrated that red cell flux varies nonlinearly with fractional flow rate. Critical flow rates, at which all or none of the cells entered one of the branches, were found to vary with diameter and hematocrit as has been reported in other studies. The data were analyzed with a theoretical model which assumes that the parent vessel contains a core of uniformly distributed red cells surrounded by a marginal gap of suspending medium; in the parent vessel lumen, the flows to the two daughter branches were assumed to be separated by a chord. The marginal gap widths and tube hematocrits deduced from the data with this model are of reasonable magnitudes.