Two models are being considered for the mechanism of chaperonin-assisted protein folding in E. coli: (i) GroEL/GroES act primarily by enclosing substrate polypeptide in a folding cage in which aggregation is prevented during folding. (ii) GroEL mediates the repetitive unfolding of misfolded polypeptides, returning them onto a productive folding track. Both models are not mutually exclusive, but studies with the polypeptide-binding domain of GroEL have suggested that unfolding is the primary mechanism, enclosure being unnecessary. Here we investigate the capacity of the isolated apical polypeptide-binding domain to functionally replace the complete GroEL/GroES system. We show that the apical domain binds aggregation-sensitive polypeptides but cannot significantly assist their refolding in vitro and fails to replace the groEL gene or to complement defects of groEL mutants in vivo. A single-ring version of GroEL cannot substitute for GroEL. These results strongly support the view that sequestration of aggregation-prone intermediates in a folding cage is an important element of the chaperonin mechanism.