SYNOPSISThe paired catchment method was used to measure the effects of riparian zone clearing and clearfelling on streamflow, stormflows and water quality. The forested riparian zone of a 39,6 ha, humid Northern Province catchment, amounting to 10 % of the catchment area, was cleared and kept clear of vegetation. Two years later the entire catchment was cleared by bulldozing and burning its cover of indigenous evergreen forest and scrub forest, and replanted to Eucalyptus grandis seedlings. The riparian clearing resulted in a small 55 mm (9 %) increase in streamflow in the first year, and mean increases per storm in quickflow volume of2,32 mm (71 %), in stormflow volume of 4,60 mm (76 %) and in storm response ratio by 2,37 % (a 50 % increase). By the second year these effects were diminished, and total flow was reduced by 56 mm (19 %) below expected flow. Clearfelling and replanting of the catchment had similarly small effects on streamflow and stormflows. This could be ascribed partly to a serious drought at the time of the treatment. Only the total streamflow could be shown to have changed significantly after the clearfelling. In the first year after clearing total flow increased by 27 mm (35 %), but by the second year the rapid establishment ofthe eucalypt crop and a dense cover of weeds had resulted in a small decrease in total flow of 12 mm (6 %). There was no significant response in total suspended sediment (TSS) concentrations to the riparian clearing. In total the sediment loss from the catchment was 0,186 tlha in the first and 0,046 tlha in the second year after the riparian treatment. In the first year after the clearfelling and afforestation of the whole catchment, the only water quality variables to change significantly were K+, NO 3' -N, PO 43,-P and Si. In the second year after clearfellingthere were changes in TSS, K+, TP, Si, N0 3'-N, and P0 43,-P. Sediment losses amounted to only 0,022 and 0,028 tlha in the first and second years respectively. None of the concentrations in stormflow showed significant changes. Most concentration levels, and especially that of sediment, were lower during the period of the drought. We conclude that: 1. indigenous evergreen forest is a relatively conservative user of water and that the quality ofwater from such catchments is high; 2, clearing of riparian and other indigenous forest and scrub vegetation is not a practical means of augmenting streamflow in this region; 3. the apparently extreme treatments applied did not disturb the soil mantle enough, or for long enough, to change the mode of streamflow generation in this catchment; 4. although the extreme clearfelling treatment had a statistically significant effect on a few water quality variables, concentrations never exceeded levels acceptable for aquatic life in river water; and 5. the minimal effects ofthe treatments may in part be due to the below average rainfall experienced during the experiment.