AbstractSoil erosion and sedimentation effects of three silvicultural treatments—clearcutting, selection cutting and no disturbance (control)—were compared in a replicated small watershed study conducted in the Ouachita Mountains of Arkansas on shallow soils derived from sandstones and shales. Clearcutting significantly increased annual sediment yields over selection and control treatments in 1981, the first year, after treatment but not in 1982 or 1983. Clearcut to control sediment yield ratios were 20:1, 6:1, and 2.6:1 in 1981, 1982, and 1983, respectively. First‐year sediment losses from clearcuts averaged 237 kg ha−1. Stream channels were stable, but they still may have been the primary source of the sediment losses. Erosion following harvest and site preparation did not exceed estimates of long‐term soil formation rates. Longterm soil losses were projected to average 70 kg ha−1 yr−1 over a 35‐yr rotation period with clearcutting while control rates were projected to average 50 kg ha−1 yr−1. A comparison of soil losses measured in this study with baseline rates and estimated soil loss tolerances suggests site productivity need not be threatened by silviculturally induced soil erosion. Suspended solids levels of stormflow were less than 100, 50, and 20 mg L−1 at least 99, 98, and 97% of the time, respectively, across treatments. Only at the 10 mg L−1 level was there a significant total suspended solids time differential in watershed stormflow between clearcut or selection cut and control treatments.