A first‐order (self‐consistent) renormalized spin‐wave theory has been used to fit the temperature dependence of the magnetizationM(T, 0) in the hexagonal ferromagnet CrBr3. However, this theory appears incapable of simultaneously explainingM(T, O) and new measurements we have made ofM(T0,H). Reexamining the first‐order theory we find that (a) self‐consistency corrections are negligible because of the low magnon density and (b) an equally satisfactory fit toM(T, O) can be obtained from a rectangular model with appropriately chosen intralayerJTand interlayerJLexchange constants and reciprocal‐lattice volume. This is possible becauseJTis so large that there is little thermal excitation beyond thekx2+ky2part of the spectrum. Since largekzmagnons are present due to the smallness ofJL, this is a strongly interacting system for which correlation effects are known to be important. Using the rectangular model we have treated spin‐wave interactions to all orders via thet‐matrix and find deviations of up to 40% from the first‐order renormalization. Since the first‐order renormalization effects onM(T, O) were shown to be large, we may infer from our calculations that the first‐order fitting of one thermodynamic function was fortuitous and was made possible by the multiparameter character of the theory.