There is presently a debate concerning the number of phases in fluorescence induction and on the identification of the several possible heterogeneities in PS II centres. However, the usual methods of analysis present numerical problems, including a lack of 'robustness' (robustness being defined as the ability to give the correct answer in the presence of distortions or artefacts). We present here the adaptation of the method of moments, which was developed for robustness, to the analysis of fluorescence induction. We were thus able to identify three phases in the fluorescence induction in the presence of DCMU. The slowest phase was attributed to the centres inactive in plastoquinone reduction by using duroquinone as electron acceptor. In order to compare fluorescence with and without DCMU, we introduced the 'rate of photochemistry', defined as the product of the area times the rate constant of an exponential. This quantity is invariant for a given centre no matter what the size of the electron acceptor pool is. The two fastest phases in the presence of DCMU were attributed to active centres because their rate of photochemistry was the same as that of the plastoquinone-reducing phases in the absence of DCMU. Because their reduction of plastoquinone showed different kinetics, these two types of active centres were either separated by more than 250 nm or were associated with discrete plastoquinone pools having restricted diffusion domains.