The kinetic behavior of the adenylyl cyclase activity associated with fat cell membranes purified by centrifugation on sucrose gradients was studied. Under most of the conditions explored, with either Mn++ or Mg++ as the divalent cation in the assay mixtures, the time courses of the reaction were not linear. In the absence of modifiers (i.e., basal activity) or in the presence of insulin, the rate tended to decrease with time; on the other hand, with fluoride or GMP-P(NH)P the curves were concave upwards. To simplify analysis of the results, two kinetic components were defined: an "initial component" corresponding to the transient rate measured between zero time and 1.5 min of assay and a "final component" corresponding to the transient rate determined between 3 and 5 min. Over the entire range of Mn++ concentration explored (0.5 to 6.0 mM), the basal initial rates were slightly higher than the final ones. With Mg++ in the range between 1.5 and 2.5 mM, the final rates were fourfold lower than the initial ones. Higher or lower Mg++ concentrations gave velocity ratios equivalent to those observed with Mn++. Insulin clearly decreased the final rates at Mn++ concentrations up to 2.5 mM. With higher concentrations the effects were completely reversed. The effects of insulin on initial rates measured with Mn++, or the initial or final rates measured with Mg++, were less evident. Stimulation of adenylyl cyclase activity by fluoride was most pronounced on the final rates. In addition, this stimulation was higher with Mg++ than with Mn++. Isoproterenol stimulation of adenylyl cyclase was negligible in the presence of Mn++ (0.5 to 6.0 mM). With Mg++ (0.5 to 6.0 mM), stimulation was more evident on the final rates. GMP-P(NH)P inhibited the initial but activated the final components of the reaction. The extent of this inhibition or activation was more pronounced with Mg++ than with Mn++. Under conditions which lead to maximal inactivation of the final component, adenylyl cyclase activity was tenfold or more higher with Mn++ than with Mg++. Similar effects were observed with GMP-P(NH)P on the initial component. However, insulin, isoproterenol and fluoride decreased the Mn++ dependence of the final component. With fluoride, the final rates measured with Mg++ were almost equivalent to those found in assays containing Mn++. Under the conditions used for measurements of adenylyl cyclase activity, the enzyme system slowly interconverts between active and inactive forms.