An on-line oxygen uptake and CO2-production rate measurement system was used to analyse the biomass activity. CO2-production rate corresponded well with oxygen uptake rate in the carbon oxidation process. In the nitrification process the ratio CO2-prod./O2-consumed (r(q)) had a larger variation, but CO2-production had a good correlation with utilized COD; variation of the r(q)-value indicate differences in the auto-/heterotrophic oxidation balance. From this basis the oxygen uptake rate for carbon oxidation and nitrification can be separated. The comparison with two biomass determination methods based on ATP-analysis shows that for heterotrophic biomass the methods indicate clearly different changes of ATP, for nitrification process the results were more compatible. By combining the biomass and on-line measurements with biomass growth model based on IAWPRC-model equations reasonably good estimates were obtained. Best fit of model was in the nitrification process; by using VSS as the measure of biomass the fit was poor. The differences in the ATP-measurement results affect also the model coefficients.An on-line oxygen uptake and CO2-production rate measurement system was used to analyze the biomass activity. CO2-production rate corresponded well with oxygen uptake rate in the carbon oxidation process. In the nitrification process the ratio CO2-prod./O2-consumed (rg) had a larger variation, but CO2-production had a good correlation with utilized COD; variation of rg-value indicate differences in auto-heterotrophic oxidation and nitrification can be separated. The comparison with two biomass determination methods based on ATP-analysis shows that for heterophil biomass the methods indicate clearly different changes of ATP, for nitrification process the results were more compatible. By combining the biomass and on-line measurements with biomass growth model based on IAWPRC-model equations reasonably good estimated were obtained. Best fit of model was in the nitrification process; by using VSS as the measure of biomass the fit was poor. The differences in the ATP-measurement results affect also the model coefficients