Several theories of the fluid-dynamic mechanism of a regenerative pump have been presented in the literature. The author has compared these theories and believes that in principle they are compatible. The differences occur in the assumptions made. Experimental data show that, although each assumption is useful for a limited operating condition and geometry, none covers a wide enough range for a complete analysis of the phenomenon. In addition, he presents a series of experimental data concerning the geometry of the pumping channel. Although the influence of the geometry of the pumping channel on performance is complicated, consistent tendencies are found. These tendencies and experimental data are available not only as an aid to design but also may serve as a guide for the improvement of any simple semiempirical theory of the regenerative pump as well as for testing any complete theories.