Metabolically stable GTP analogues were 10 to 40 times more potent activators of DL-isoproterenol-stimulated adenylate cyclase (EC 4.6.1.1) from pigeon erythrocyte membranes that GTP. The order of effectiveness was guanosine-5'-O-(3-thiotriphosphate) (GTP gamma S) greater than guanylyl imidodiphosphate (Gpp(NH)p) greater than guanylyl methylenediphosphonate (Gpp(CH2)p greater than GTP. In contrast to activation with GTP, activation by analogues was independent of ATP concentration. The analogues seem to bind, however, to the same regulatory sites in membrane preparations to which GTP is bound but with higher affinity; Kdiss for (14C)Gpp-(nh)p and (3H)Gpp(CH2)p and membranes was 0.7 and 2.4 x 10-7 M, respectively. DL-Isoproterenol did not increase the amount of guanylnucleotide bound, it merely accelerated and potentiated activation. Bound radioactive GTP analogues were recovered unchanged from the membrane pellet. This and mutual displacement of analogues and GTP ruled out covalent attachment of the whole or of part of the nonphosphorylating GTP analogues. Treatment of the membrane preparation with Gpp(NH)p effectively (greater than 80%) protected DL-isoproterenol-activated adenylate cyclase against the action of Filipin or Lubrol PX. Activation of membrane-bound adenylate cyclase with GTP analogues resulted in a stable enzyme which could be nearly completely resolved from membranes with Lubrol PX and stripped of lipids and detergent without loss of activity. This effect was synergistically amplified by DL-isoproterenol. A protein fraction with an apparent molecular weight of 230,000, containing about 90% of (14C)Gpp(NH)p originally bound to membranes, could be solubilized and separated from adenylate cyclase activity by chromatography on Sepharose 4B. The binding protein was purified about 40- to 80-fold from activated membranes. Removal of the nucleotide binding protein was also achieved by affinity chromatography with GTP gamma S coupled to Sepharose via a spacer. When membranes which were not or only weakly and reversibly activated (with GMP) were used as source of the soluble preparation, removal of the binding protein resulted in 75% loss of Gpp(NH)p activation without change in basal and Mg2+/F-stimulated adenylate cyclase activity. It is assumed that the GTP analogues cause an unphysiological, irreversible activation of membrane-bound adenylate cyclase, because, in contrast to the natural guanylnucleotides whose action they mimic, they are metabolically inert and bound quasi-irreversibly to regulatory sites.