The hypothesis of ‘localized’ chemiosmosis have been put to an experimental test in the system of oxidative phosphorylation by rat liver mitochondria. We find that the variation of the ratio of phosphate potential to delta muH with delta muH does not depend on how delta muH is varied. This is in conflict with hypotheses of localized chemiosmosis. Of all coupling hypotheses, only the parallel-coupling hypothesis can explain the observations, unless variation of the H+/ATP stoichiometry of the ATPase proton pump is accepted. In the latter case. ‘macroscopic’ chemiosmosis can explain the observations equally well as hypotheses of localized chemiosmosis. It is concluded that either variation of H+/ATP stoichiometries must be accepted, or that the parallel-coupling hypothesis should be reformulated so that it becomes open to experimental tests.