The mechanism of erythrocyte settling has been observed directly by microscope. The results have been correlated with the standard method of measuring erythrocyte sedimentation rate (E.S.R.) and with theories of the settling of suspensions. Downward movement of cells and aggregates causes rapid upward currents in the suspending medium carrying with them other cells and smaller aggregates that may collide and result in attachment or break-up. The fall of the interface at the top of the cell column has three phases: (a) an initial acceleration, (b) a long phase at a constant rate of fall and (c) a deceleration phase as the aggregates pack at the bottom of the tube. During phases a and b, the aggregate size at the top of the column is constantly increasing to reach saturation when the interface meets the pile-up of cells at the bottom of the chamber, when phase c starts. The results for cells suspended in Ringer's solution, in which there is no aggregation, are compatible with the theory of "hindered settling" of slurrys in the chemical engineering literature. Settling in plasma is complicated by the formation and break-up of aggregates during settling, and will require a more elaborate theory.