The rate of release of bound c[3H]AMP from the two types (A and B) of cAMP binding sites on the regulatory subunit dimer (R2I) of rabbit muscle protein kinase I was studied in the presence of the catalytic (C) subunit of protein kinase. Rebinding of released c[3H]AMP was avoided by using highly diluted reactants or adding unlabeled cAMP or its analogues. No significant C-induced dissociation of R2I-(c[3H]AMP)4 occurred in the absence of Mg2+-ATP. Of the two options that one or two molecules of C are required to induce the release of c[3H]AMP bound to R2I, only the first one was compatible with the first-order dependence on [C] of the rate of release of c[3H]AMP observed over a wide range of C concentrations. In the absence of added unlabeled cyclic nucleotide, the rate of the C-induced release of c[3H]AMP was the same from site A and site B. The apparent second-order rate constant for the association of C to R2I(c[3H]AMP)4 was 6 X 10(6) M-1 s-1 (37 degrees C, 0.15 M KCl). Raising the concentration of unlabeled cAMP in the medium up to 1 microM decreased by up to 50% the rate of the C-induced release of bound c[3H]AMP from both sites. This is explained by assuming that the association of one molecule of C to R2I(c-[3H]AMP)4 leads to the release of c[3H]AMP first from one R subunit and subsequently, by a process that can be blocked by about 1 microM cAMP, from the other R subunit. A further rise of the cAMP concentration decreased the rate of release from site B only, so that the C-induced release of c[3H]AMP occurred almost exclusively from site A at very high concentrations of cAMP. This suggests that c[3H]AMP is released first from site A and that this vacant site by interacting with cAMP inhibits the release of c[3H]AMP from site B of the same R subunit. The role of site A in controlling the C-induced release was further supported by the finding that several cAMP analogues inhibited the release with potencies correlating with their affinities for site A. The C-induced release of c[3H]AMP from aged R2I was about 10 times slower than that from fresh R2I. No significant C-induced release of c[3H]AMP was observed from the monomeric fragment obtained by limited trypsin treatment of R2(1).