Upon onset of saturating continuous light only the first part of the observed polyphasic fluorescence rise follows Q(A) reduction (photochemical phase), whereas the remaining part (thermal phases) is kinetically limited by relatively slow reactions with light saturated half-times in the order of 10-50 ms. A simple hypothesis is presented for the interpretation of these fundamentally different types of variable fluorescence. The hypothesis, which is based on the reversible radical pair model of PSII, assumes stimulation of both prompt and recombination fluorescence upon Q(A) reduction, with only recombination fluorescence being in competition with nonradiative energy loss processes at the reaction centers. It is proposed that changes in the rate constants of these processes modulate the yield of recombination fluorescence in closed centers, thus causing large variations in the maximal fluorescence yield and also giving rise to the 'thermal phases'. This hypothesis can reconcile numerous experimental findings which so far have seemed difficult to interpret.