Two lab-scale plug flow activated sludge reactors were run in parallel for 4 months at 30 and 55°C. Research focussed on: (1) COD (chemical oxygen demand) removal, (2) effluent turbidity at both temperatures, (3) the origin of effluent colloidal material and (4) the possible role of protozoa on turbidity levels. Total COD removal percentages over the whole experimental period were 66-7t 30°C and 53-11t 55°C. Differences in total COD removal between both systems were due to less removal of soluble and colloidal COD at 55°C compared to the reference system. Thermophilic effluent turbidity was caused by a combination of influent colloidal particles that were not effectively retained in the sludge flocs, and erosion of the thermophilic activated sludge itself, as shown by denaturing gradient gel electrophoresis (DGGE) profiles. DGGE analysis of PCR-amplified 16S rDNA fragments from mesophilic and thermophilic sludge differed, indicating that different microbial communities were present in the two reactor systems. The effects of protozoal grazing on the effluent turbidity of both reactors was negligible and thus could not account for the large turbidity differences observed.