Although the adenylate cyclase of Dictyostelium discoideum cannot be activated by its cAMP agonist in vitro, its in vivo activation can be demonstrated by rapidly breaking and assaying the cells, over 10-fold higher activity being observed for stimulated cells than for basal cells. We report here that when basal cells are broken in the presence of labeled ATP and then rapidly assayed, they display 8-fold more adenylate cyclase activity than cells broken in the presence of unlabeled ATP. This suggests that a significant amount of the enzyme in extracts of basal cells is sequestered within vesicles that can be loaded with substrate at the time of cell lysis, but then rapidly seal. In contrast to the results obtained with basal cells, when cells activated in vivo are broken in the presence of labeled ATP, there is less than 2-fold increase in adenylate cyclase activity. Thus, a much smaller percentage of the observed adenylate cyclase activity of stimulated cells appears to be due to sequestered enzyme than of basal cells. Two models are discussed that account for these observations. One model envisions that roughly equal populations of sequestered and nonsequestered enzyme are produced upon breakage of both basal and activated cells, but that sequestered enzyme in basal extracts becomes uniquely activated in vitro. The other model proposes that the differences in observed activity are due directly to differences in sequestration. According to this latter model, nearly all of the -fold activation previously observed for the D. discoideum adenylate cyclase can be accounted for by a change in sequestration of the enzyme rather than by an intrinsic alteration in the enzyme per se. It therefore suggests a novel mode of regulation whereby an enzyme may be packaged within vesicles and its activity controlled by modulating the permeability of the vesicles to its substrate or effectors.