Quantitative fluorescence in situ hybridization (FISH) and the combination of FISH with microautoradiography (MAR) were used in order to study the long-term population dynamics (2.5 years) and the in situ physiology in two parallel activated sludge pilot systems with enhanced biological phosphorus removal (EBPR). The two systems received the same influent wastewater, but were differently operated (with and without nitrogen removal, respectively). Both systems showed a significant P removal that increased when different substrates (phosphorus (P), acetate and glucose, respectively) were added to the influent wastewater. Rhodocyclus-related bacteria were present in both systems in significant numbers (ranging from 4 to 28%) throughout the whole period. This supports the hypothesis that these bacteria occur in significant numbers in different types of well-operating EBPR activated sludge processes. However, we observed a lower correlation (< 0.5) for the amount of Rhodocyclus-related bacteria to the P content in activated sludge than previous studies (> 0.9). The Actinobacteria were the only additional group of bacteria which showed a similar degree of correlation to the P content in activated sludge as the Rhodocyclus-related bacteria--but only for the system without nitrogen removal. Significant amounts (< or = 12%) of glycogen-accumulating bacteria (GAOs) were detected in the system with nitrogen removal (but not in the other system), but had no, in contrast to previous observations, apparent negative effect on the overall EBPR performance. FISH-MAR indicated that a significant part of the Betaproteobacteria (part of them identified as Rhodocyclus-related bacteria) as well as the Actinobacteria were able to take up 33Pi, [3H]-acetate and [3H]-glucose under anaerobic-aerobic conditions. The contribution of anoxic 33Pi uptake under alternating anaerobic-anoxic conditions was significantly lower. Interestingly, not all Rhodocyclus-related bacteria showed uptake of these three radioactive substrates. This may be due to differences in metabolic state, physiological potential or genotype, not detectable by the present probe set for Rhodocyclus-related bacteria. Comparison of the 33Pi, [3H]-acetate and [3H]-glucose uptake by activated sludge after different fixation and incubation procedures showed that a part of the observed 33Pi uptake may have been caused by a combination of a biological and chemical or biologically induced chemical P adsorption.