Photoinhibition of photosynthesis in willow (Salix sp.) leaves was investigated by measuring the ratio of variable (Fv) to maximal (FM) chlorophyll fluorescence. The use of this parameter was justified as it showed similar kinetics of recovery from photoinhibition as did the light-limited rate of gross photosynthesis. When leaves of different status were exposed to different environmental conditions for 1 h, the large variation in photoinhibition subsequently observed was not correlated with the incident photosynthetic photon flux density. However, it was positively correlated with the fraction of closed reaction centres of photosystem (PS) II, and negatively correlated with the magnitude of the high-energy fluorescence quenching, both of which were measured during the treatment. The best correlation was obtained when the two parameters were taken together in a multiple regression model. Young high-light-acclimated leaves were the most resistant to photoinhibition at a given fraction of closed PSII centres, which was explained by these having the most prominent high-energy quenching. Leaves illuminated in the presence of dithiothreitol and the uncoupler nigericin were found at the other end of the correlation. However, the markedly lowered high-energy quenching induced by dithiothreitol was not fully matched by the expected increase in photoinhibition. The fraction of closed PSII centres provided a relative not an absolute measure of excessive excitation, since the correlation model varied depending on the length of the treatment. In a separate experiment the response to fluctuating light under otherwise constant conditions was studied. It was found that under these more restricted conditions photoinhibition could be predicted from the integrated light dose.