We have investigated the properties of the cytoplasmic domain (FlhB(C)) of the 383-amino-acid Salmonella membrane protein FlhB, a component of the type III flagellar export apparatus. FlhB, along with the hook-length control protein FliK, mediates the switching of export specificity from rod- and hook-type substrates to filament-type substrates during flagellar morphogenesis. Wild-type FlhB(C) was unstable (half-life, ca. 5 min), being specifically cleaved at Pro-270 into two polypeptides, FlhB(CN) and FlhB(CC), which retained the ability to interact with each other after cleavage. Full-length wild-type FlhB was also subject to cleavage. Coproduction of the cleavage products, FlhB(delta CC) (i.e., the N-terminal transmembrane domain FlhB(TM) plus FlhB(CN)) and FlhB(CC), resulted in restoration of both motility and flagellar protein export to an flhB mutant host, indicating that the two polypeptides were capable of productive association. Mutant FlhB proteins that can undergo switching of substrate specificity even in the absence of FliK were much more resistant to cleavage (half-lives, 20 to 60 min). The cleavage products of wild-type FlhB(C), existing as a FlhB(CN)-FlhB(CC) complex on an affinity blot membrane, bound the rod- and hook-type substrate FlgD more strongly than the filament-type substrate FliC. In contrast, the intact form of FlhB(C) (mutant or wild type) or the FlhB(CC) polypeptide alone bound FlgD and FliC to about the same extent. FlhB(CN) by itself did not bind substrates appreciably. We propose that FlhB(C) has two substrate specificity states and that a conformational change, mediated by the interaction between FlhB(CN) and FlhB(CC), is responsible for the specificity switching process. FliK itself is an export substrate; its binding properties for FlhB(C) resemble those of FlgD and do not provide any evidence for a physical interaction beyond that of the export process.