A number of potential models for the interaction of cyclic AMP with protein kinase (RC or R2C2) have been examined. These include: Model 1, the simultaneous binding of cyclic AMP and release of C (catalytic subunit) from an independent RC protomer; Model 2, dissociation of an independent RC protomer prior to cyclic AMP binding to R (regulatory subunit); Model 3, cyclic AMP binding to RC prior to the dissociation of C; Model 4, random binding of cyclic AMP and dissociation of C with an interaction factor alpha less than 1; Model 5, release of 2C concomitant with the binding of one cyclic AMP to R2C2 followed by binding of the second cyclic AMP to the vacant R subunit; and Model 6, the simultaneous binding of cyclic AMP and release of C from one RC protomer resulting in a greater "affinity" of the other RC protomer for cyclic AMP, i.e., a cooperative version of Model 1. All the above models yield [cyclic AMP]0.5 values that increase with increasing protein concentration and Hill plots with average slopes equal to or less than 1.0 in the usual experimental range (10 to 90% of saturation). The Hill plots can be nonlinear, but for each model the exact shape of the plot changes in a characteristic (diagnostic) manner with changing protein concentration. Skeletal muscle protein kinase yields relatively linear Hill plots with napp values greater than 1.0. Consequently, Models 1 to 6 are not likely candidates. However, Model 2 is an excellent alternative model for proteins that display "negative cooperativity" with respect to the binding of a ligand. The properties of several "linear", "tetrahedral", and "all-or-nothing" cooperative models have also been examined. These include Models 7, A, B, and C and 8, A, B, and C which are cooperative versions of Models 2 and 3, respectively, and Model 9, a cooperative version of random Model 4. Model 9 is the most general model from which all others can be derived. Models 9 and 7, A, B, and C in which the prior dissociation of C greatly enhances or is an absolute requirement for cyclic AMP binding to R, are likely candidates for skeletal muscle protein kinase. All four of these models are capable of yielding Hill plots with average slopes greater than 1, and napp values that decrease with increasing protein concentration (in agreement with published data). In addition, in all four models the tight binding of MgATP to R2C2 yields decreased napp values and increased [cyclic AMP]0.5 values (also consistent with published data).