1. When magnesium and orthophosphate are added to Na+,K+-ATPase containing occluded rubidium ions, and suspended in a medium containing free rubidium ions, only 50% of the occluded rubidium is released rapidly. This is because the release of occluded rubidium is ordered, and the replacement (by rubidium ions from the medium) of the first occluded rubidium ions to leave slows the departure of the remaining occluded ions. 2. Since the Na+,K+-ATPase probably exists in the membrane as a structural dimer, the ordered release might represent either the ordered emptying of the two halves of the dimer, or the ordered release of the two rubidium ions thought to be contained in each promoter. 3. The present experiments were designed to decide between these possibilities by examining the behaviour of Na+,K+-ATPase in which about half of the protomers had been randomly inactivated by pre-treatment either with fluorescein isothiocyanate or with alpha-chymotrypsin. 4. The results show that the release of rubidium ions from each protomer is ordered.