This research set out to investigate reported anomalies in the calculation of COD balances in biological nutrient (nitrogen and phosphorus) removal (BNR) activated sludge systems. For non-BNR systems, accurate COD balances are consistently attainable from experimental measurements. That is, influent COD can be accounted for in the effluent flow, waste sludge stream, and mass of oxygen utilised for carbonaceous oxidation. For BNR systems, in a number of instances where COD balances have been performed, the balances do not close. That is, the sum of the COD leaving the system is as much as 20% less than the COD entering the system, and thus there is an apparent COD 'loss'. In attempting to explain the problem, a laboratory based experimental program was developed which isolated specific areas of interest within BNR systems. The study involved two main experimental phases. In the first experimental phase, a closed denitrification assay technique was developed. This assay technique allowed specific aspects which have relevance to be evaluated: the nitrate-to-oxygen conversion factor, and the yield of activated sludge organisms under anoxic conditions. The advantages of the closed assay were that all of the needed COD balance terms were independently measurable, and the denitrification system was sealed which isolated the assay environment from the surrounding environment and minimised the effects of external inputs (i.e. oxygen transfer from the air to the liquid). In the second experimental phase, an excess biological phosphorus removal (EBPR) system was operated to investigate the influence of the influent COD to phosphorus ratio on EBPR biological activity and COD balance calculations. The EBPR system was a laboratory - scale sequenching batch reactor (SBR) which was extensively monitored over an eight month period. This allowed for the investigation of several factors such as, the rate of anaerobic phosphorus release and COD uptake, the ratio of anaerobic phosphorus released to COD taken up, and the rate of aerobic phosphorus uptake. The system also allowed for the calculation of COD balances. The main body of this thesis is presented as a series of five papers. The first aper (Chapter 3) presents a study on the nitrate-to-oxygen conversion factor for denitrification which theoretically is 2.86 gO₂/gNO₃-N. That is, when nitrate replaces oxygen as electron acceptor, the mass of nitrate reduced can be converted to oxygen equivalence - for purposes of COD balancing - through the use of this conversion factor. Calculations from a series of denitrifying experiments resulted in an observed conversion factor of 2.96 gO₂/gNO₃-N. However, the observed factor could not be distinguised statistically from the theoretical value, hence confirming the theoretical factor. The second paper presents a study of sludge production under anoxic (denitrifying) and aerobic conditions. Decreased sludge production and COD 'losses' are confounded in nutrient removal activated sludge systems. Hence this portion of the investigation was initiated to separate these two influences. Batch tests were performed under anoxic (denitrifying) and aerobic conditions using various organisms. The results show that COD balances were achieved, but differences in sludge production under the two conditions indicate a difference in true yield between the environments. An anoxic yield of 0.402 mg particulate COD/mg consumed COD was determined and compared with an observed aerobic yield of 0.645 mg particulate COD/mg consumed COD. These results dispel the assertion that less sludge production results because of COD 'losses'. That is, decreased sludge production in BNR systems can at least in part be explained by a lower yield during unaerated periods. The third and fourth papers (Chapters 5 & 6) present the results from the EBPR system. The third paper conccentrates on the results from the SBR and the influence of the various influent COD to phosphorus ratios. Five different influent ratios were investigated ranging from 8 to 98 (mgCOD/mgP) using a synthetic feed with acetate as the sole carbon source. The phosphorus content of the waste sludge increased from 4 to 17 per cent of the total solids as the influent COD:P ratio decreased. However, complete anaerobic COD uptake was observed irrespective of the phosphorus content of the sludge. Also, COD balances on the reactor averaged 1.04 indicating no apparent COD 'loss' in the system. The fourth paper presents the results of batch tests performed on the waste sludge from the SBR. Anaerobic batch tests were designed to determine the influence of the SBR steady state influent COD to phosphorus ratio on the kinetics and stoichiometry of the system's microbial community. In particular, the rate of substrate uptake increased from 104 to 211 mgCOD/gVSS/hr as the influent ratio decreased. Similarly, the rate of phosphorus release increased from 17 to 166 mgP/gVSS/hr as the influent ratio decreased. The batch test results showed that the ratio of COD taken up to phosphorus released under anaerobic conditions varies with the phosphorus content of the sludge. The observed release ratio increased from 0.17 to 0.79 mgP/mgCOD as the phosphorus content increased. The final paper (Chapter 7) presents a theoretical biochemical model for EBPR systems. The proposed model makes a distinction between glycogen accumulating organisms (GAOs) and polyphosphate accumulating organisms (PAOs) but suggests that significant populations of both types of organisms exist in most EBPR systems. That is, it is proposed that only in systems that are stressed does one type of organism dominate the mixed community. The biochemical models of Comeau et al. (1986) and Wentzel et al. (1986) were merged with the biochemical model of Satoh et al. (1994) to form a comprehensive model for EBPR system activity. The two-organism model provides several advantages over single organism models including the ability to predict variable rates of reaction and a variable anaerobic COD uptake to phosphorus release ratio.