The process of protein folding in the cell is now known to depend on the action of other proteins. These proteins include molecular chaperones, which interact non-covalently with proteins as they fold and improve the final yields of active protein in the cell. The precise mechanism by which molecular chaperones act is obscure. Experiments reported recently show that for one molecular chaperone (Cpn60, typified by the E. coli protein GroEL), the folding reaction is driven by cycles of binding and release of the co-chaperone Cpn10 (known as GroES in E. coli). These alternate with binding and release of the unfolded protein substrate. These cycles come about because of the opposite effects of Cpn10 and unfolded protein on the Cpn60 complex: the former stabilises the ADP-bound state of Cpn60, whereas the latter stimulates ADP-ATP exchange. This model proposes that the substrate protein goes through multiple cycles of binding and release, and is released into the cavity of the Cpn60 complex where it can undergo folding without interacting with other nearby folding intermediates. This is consistent with the ability of Cpn60 proteins to enhance folding by blocking pathways to aggregation.