The effect of influent loading rate on mass removal of BOD, SS and faecal coliforms (FC) from dairy parlour wastewaters was compared in four pairs of planted (Schoenoplectus validus) and unplanted gravel-bed wetlands (each 19 m 2 ). The wetlands were operated at nominal retention times of 7, 5.5, 3 and 2 days, with in and outflows sampled fortnightly over a 20 month period. Hydraulic flows were monitored to enable calculation of the mass flows of pollutants. Influent water quality varied markedly over the trial period (CBOD 5 20-300 g m - 3 , SS, 60-250 g m - 3 ; FC, 10 3 -10 6 MPN (100 ml - 1 ). NBOD was an important component of total BOD, being around 1.5 times higher than the influent CBOD 5 , and 2-10 times higher than the effluent CBOD 5 . Outflow levels of CBOD 5 , SS and faecal coliforms rapidly mirrored changes in influent loadings. Mean mass removal of CBOD 5 increased from 60-75% to 85-90%, total BOD (CBOD 5 + NBOD) from 50 to 80% and FC from 90-95 to > 99% with increasing wetland retention time during the first 12 months of monitoring. Mean annual SS removals of 75-85% were recorded irrespective of loading rate. High levels of dissolved humic colour in the wastewaters were little affected by passage through the wetland at short retention times, but were reduced by up to 40% at longer retentions. Mass removals of CBOD 5 , SS and FC showed monotonic relationships to mass loading rates, with little difference between the performance of planted and unplanted wetlands, except for CBOD 5 at high loadings (> 3 g m - 2 d - 1 ). The planted wetlands showed significantly improved removal rates for CBOD 5 at higher loadings, and 1.3 to 2.6 fold higher mass removals of total BOD.