The effect of GDP on rabbit heart adenylate cyclase has been determined under conditions where only 0.08% to 0.26% of an added 100 microM was converted to GTP in the course of the assay. At concentrations of 100 microM, GDP stimulated basal cyclase activity to the same extent as GTP and guanosine-5'-O-(2-thiodiphosphate) (GDP beta S). Isoproterenol increased activity in the presence of GTP or guanylyl-imidodiphosphate (Gpp(NH)p), but not in the presence of GDP or GDP beta S. It is suggested that the hydrolysis of GTP to GDP is the "turn-off" mechanism for beta-receptor stimulation of cardiac adenylate cyclase, but not for stimulation by GTP alone. The effects of GDP and GDP beta S are readily removed by washing, implying that their binding to Ns (the guanine nucleotide binding protein) is weak. GDP beta S initially competes with Gpp(NH)p, reducing Gpp(NH)p-stimulated activity. As stimulation of cyclase activity by Gpp(NH)p develops, in the course of 30 min, Gpp(NH)p becomes no longer displaceable by GDP beta S. Isoproterenol does not release 3H-Gpp(NH)p or reduce Gpp(NH)p-stimulated activity, once the nucleotide has become tightly bound. Nor does isoproterenol change the relative affinities of GDP beta S and Gpp(NH)p when these analogs are given together. There is, therefore, no evidence that isoproterenol acts by releasing tightly bound GDP from Ns, or that it 'unlocks' the guanine nucleotide binding site in the myocardial sarcolemma. In this, the cardiac adenylate cyclase system differs from the avian erythrocyte system. The action of isoproterenol is best explained by an increased dissociation of alpha(GTP) and beta,gamma-subunits of the Ns protein.