Illumination of photosystem II by a saturating short flash results in a stabilized charge separation in only about 90% of the reaction centers. During a series of flashes, the 10% fraction of "photochemical misses" is randomly redistributed among the centers. This phenomenon is investigated in DCMU-inhibited material, eliminating the contribution to misses due to electron transfer equilibrium on the quinone acceptors. Under such conditions, the miss coefficient is about 5% and is enhanced to about 40% in the presence of hydroxylamine at low pH. When a second flash is fired, its efficiency increases as a function of the time delay after the first flash (turnover experiments). This process involves three distinct time domains: <10 micros, 100 micros, and 10 ms. From a study of the 515-nm field-indicating change, it appears that the increased inefficiency caused by hydroxylamine is not due to a lesser amount of initial charge separation but to a recombination process concomitant with the 100 micros phase of the turnover. The slow turnover phase (10 ms) is not associated with a recombination or any other electron transfer event but reflects a modification of open centers during which their probability to achieve charge stabilization rather than recombination is progressively increased. These results are interpreted in terms of an equilibrium between two conformational states of the centers endowed with different stabilization yield ("good" and "bad " stabilizers). The 100 micros turnover phase is due to the reopening of the bad stabilizers by recombination, and the 10 ms phase accompanies the redistribution of these centers among the two conformational states.