The results of a numerical study as to the adequacy of the point kinetics method for analyzing rod-ejection accidents are presented. Two-group slab representations of three different seed-blanket reactors are considered. Five different point kinetics formulations are considered for each fictitious rod-ejection accident considered; each formulation being characterized by the shape functions used to calculate the instantaneous reactivity. From these numerical studies we conclude that the accuracy of a point model, for rapid, nonseparable transients of the type considered, is extremely dependent upon the specifics of the particular model (i.e., on the shape function used), and in general one cannot assume a priori that a specific point model is conservative or nonconservative.