An 11-year paired-catchment study was implemented to assess the effect of differing logging practices on streamflow in mixed-species eucalypt forest at Tantawangalo Creek, southeastern Australia. Following a 3-year calibration period, Ceb catchment was retained as a control, Wicksend catchment was patch-cut to remove 22% of basal area, and Willbob catchment was thinned to remove 12% of basal area. The treatments resulted in an initial increase in monthly total streamflow and baseflow at both treated catchments. In annual terms, total streamflow increased by 10% in the first 3 years after logging at Wicksend, and by 31% for the first 4 years at Willbob, despite the difference in basal area removed. Streamflow then returned to pre-treatment levels for 1 year and subsequently declined by 20% below pre-treatment levels over the next 4 years at Wicksend. Willbob displayed a return to, but not below, pre-treatment levels over the same period. Although there were some large post-treatment stormflows, changes in baseflows were largely responsible for the streamflow increases, and at Wicksend, explained the streamflow decreases in the latter observation period. Differences in regeneration densities between the treated catchments indicate the greater magnitude and persistence of flow increases at Willbob were due to poor regeneration following the thinning treatment, while dense and vigorous under- and overstorey regeneration dampened initial flow increases and facilitated significant decreases at Wicksend. Significant streamflow decreases after disturbance have not been widely reported for mixed-species eucalypt forests, and the magnitude of streamflow increases after only 12% basal area removal at Willbob is unusual.