Its superior quantum efficiency renders PSII a model for biomimetic systems. However, also in biological water oxidation by PSII, the efficiency is restricted by recombination losses. By laser-flash illumination, the secondary radical pair, P680+Q−A (where P680 is the primary Chl donor in PSII and QA, primary quinone acceptor of PSII), was formed in close to 100% of the PSII. Investigation of the quantum efficiency (or yield) of the subsequent steps by time-resolved delayed (10 μs to 60 ms) and prompt (70 μs to 700 ms) Chl fluorescence measurements on PSII membrane particles suggests that (1) the effective rate for P680+ Q−A recombination is approximately 5 ms−1 with an activation energy of approximately 0.34 eV, circumstantially confirming dominating losses by reformation of the primary radical pair followed by ground-state recombination. (2) Because of compensatory influences on recombination and forward reactions, the efficiency is only weakly temperature dependent. (3) Recombination losses are several-fold enhanced at lower pH. (4) Calculation based on delayed-fluorescence data suggests that the losses depend on the state of the water-oxidizing manganese complex, being low in the S0→S1 and S1→S2 transition, clearly higher in S2→S3 and S3→S4⇒S0. (5) For the used artificial electron acceptor, the efficiency is limited by acceptor-side processes/S-state decay at high/low photon-absorption rates resulting in optimal efficiency at surprisingly low rates of approximately 0.15–15 photons s−1 (per PSII). The pH and S-state dependence can be rationalized by the basic model of alternate electron-proton removal proposed elsewhere. A physiological function of the recombination losses could be limitation of the lifetime of the reactive donor-side tyrosine radical (Y·Z) in the case of low-pH blockage of water oxidation.