Hsp70 chaperones are heterotropic allosteric systems in which ATP and misfolded or aggregated polypeptides are the activating ligands. To gain insight into the mechanism by which ATP and polypeptides regulate Hsp70 chaperone activity, the effect of a short peptide on the K(M) for ATP was analyzed using the Escherichia coli Hsp70 called DnaK. In the absence of peptide, the K(-P)(M) for ATP is 52 +/- 11 nM, whereas this value jumps to 14.6 +/- 1.6 microM in the presence of saturating peptide. This finding supports a mechanism in which ATP binding drives the chaperone in one direction and peptide binding pushes the chaperone back in the opposite direction (and thus increases K(M)), according to ATP + DnaK.P <==> ATP.DnaK.P <==> ATP.DnaK* + P, where ATP.DnaK.P is an intermediate from which competing ATP hydrolysis occurs (ATP.DnaK.P --> ADP.DnaK.P). We show that this branched mechanism can even explain how DnaK hydrolyzes ATP in the absence of peptide and that the true rate constant for DnaK-mediated ATP hydrolysis (k(hy)) in the absence of peptide may be as high as 0.5 s(-)(1) (rather than 5 x 10(-)(4) s(-)(1) as often stated in the literature). What happens is that a conformational equilibrium outcompetes ATP hydrolysis and effectively reduces the concentration of the intermediate by a factor of a thousand, resulting in the following relation: k(cat) = k(hy)/1000 = 5 x 10(-)(4) s(-)(1). How polypeptide substrates and the co-chaperone DnaJ modulate DnaK to achieve its theoretical maximal rate of ATP hydrolysis, which we suggest is 0.5 s(-)(1), is discussed.