Isolated chloroplast coupling factor 1 ATPase is known to retain about 1 mol of tightly bound ADP/mol of enzyme. Some experimental results have given evidence that the bound ADP is at catalytic sites, but this view has not been supported by observations of a slow replacement of the bound ADP when CaATP or MgATP is added. The experiments reported in this paper show why a slow replacement of ADP bound at a catalytic site can occur. When coupling factor 1, labeled with tightly bound [3H]ADP, is exposed to Mg2+ or Ca2+ prior to the addition of MgATP or CaATP, a pronounced lag in the onset of ATP hydrolysis is observed, and only slow replacement of the [3H]ADP occurs. Mg2+ or Ca2+ can induce inhibition very rapidly, as if an inhibited form of the enzyme results whenever the enzyme with tightly bound ADP encounters Mg2+ or Ca2+ prior to ATP. The inhibited form can be slowly reactivated by incubation with EDTA, although some irreversible loss in activity is encountered. In contrast, when MgATP or CaATP is added to enzyme depleted of Mg2+ and Ca2+ by incubation with EDTA, a rapid onset of ATP hydrolysis occurs and most of the tightly bound [3H]ADP is released within a few seconds, as expected for binding at a catalytic site. The Mg2+-induced inhibition of both the ATPase activity and the lack of replacement of tightly bound [3H] ADP can be largely prevented by incubation with Pi under conditions favoring Pi addition to the site containing the tightly bound ADP. Our and other results can be explained if enzyme catalysis is greatly hindered when MgADP or CaADP without accompanying Pi is tightly bound at one of the three catalytic sites on the enzyme in a high affinity conformation.