Fluoride ion, presumably an Al3+-F- complex, has been proposed to activate the guanine nucleotide regulatory protein (G-protein) of the visual system, transducin, by associating with GDP at the nucleotide-binding site and thus mimicking the effects of non-hydrolysable GTP analogues [Bigay, Deterre, Pfister & Chabre (1985) FEBS Lett. 191, 181-85]. We have examined this proposed model by using the adenylate cyclase complexes of frog erythrocytes, S49 lymphoma cells and human platelets. Preincubation of plasma membranes from frog erythrocytes and S49 cells with 20 mM-fluoride for 20 min at 30 degrees C strongly stimulated adenylate cyclase activity. In contrast, the preactivated membranes were still able to bind beta-adrenergic agonist with high affinity, as determined by radioligand-binding techniques. Moreover, high-affinity agonist binding in fluoride-treated membranes was fully sensitive to guanine nucleotide, which decreased beta-adrenergic-receptor affinity for agonist. Very similar results were obtained for [3H]prostaglandin E1 binding to S49 membranes pretreated with fluoride. Incubation of human platelet membranes with increasing concentrations of fluoride (1-50 mM) resulted in biphasic regulation of adenylate cyclase activity, with inhibition observed at concentrations greater than 10 mM. Preincubation of platelet membranes with 20 mM-fluoride did not affect agonist high-affinity binding to alpha 2-adrenergic receptors, nor receptor regulation by guanine nucleotide. These results suggest that the model developed from the study of transducin may not be generally applicable to the G-proteins of the adenylate cyclase system.