Watershed mass balances for solutes of atmospheric origin may be complicated by the residence times of water and solutes at various time scales. In two small forested headwater catchments in the Appalachian Mountains of Virginia, USA, mean annual export rates of SO₄⁼ differ by a factor of 2, and seasonal variations in SO₄⁼ concentrations in atmospheric deposition and stream water are out of phase. These features were investigated by comparing ³H, ³⁵S, δ³⁴S, δ²H, δ¹⁸O, δ³He, CFC-12, SF₆, and chemical analyses of open deposition, throughfall, stream water, and spring water. The concentrations of SO₄⁼ and radioactive ³⁵S were about twice as high in throughfall as in open deposition, but the weighted composite values of ³⁵S/S (11.1 and 12.1 × 10^− 15) and δ³⁴S (+ 3.8 and + 4.1‰) were similar. In both streams (Shelter Run, Mill Run), ³H concentrations and δ³⁴S values during high flow were similar to those of modern deposition, δ²H and δ¹⁸O values exhibited damped seasonal variations, and ³⁵S/S ratios (0–3 × 10^− 15) were low throughout the year, indicating inter-seasonal to inter-annual storage and release of atmospheric SO₄⁼ in both watersheds. In the Mill Run watershed, ³H concentrations in stream base flow (10–13 TU) were consistent with relatively young groundwater discharge, most δ³⁴S values were approximately the same as the modern atmospheric deposition values, and the annual export rate of SO₄⁼ was equal to or slightly greater than the modern deposition rate. In the Shelter Run watershed, ³H concentrations in stream base flow (1–3 TU) indicate that much of the discharging ground water had been deposited prior to the onset of atmospheric nuclear bomb testing in the 1950s, base flow δ³⁴S values (+ 1.6‰) were significantly lower than the modern deposition values, and the annual export rate of SO₄⁼ was less than the modern deposition rate. Concentrations of ³H and ³⁵S in Shelter Run base flow, and of ³H, ³He, CFC-12, SF₆, and ³⁵S in a spring discharging to Shelter Run, all were consistent with a bimodal distribution of discharging ground-water ages with approximately 5–20% less than a few years old and 75–95% more than 40 years old. These results provide evidence for 3 important time-scales of SO₄⁼ transport through the watersheds: (1) short-term (weekly to monthly) storage and release of dry deposition in the forest canopy between precipitation events; (2) mid-term (seasonal to interannual) cycles in net storage in the near-surface environment, and (3) long-term (decadal to centennial) storage in deep ground water that appears to be related to relatively low SO₄⁼ concentrations in spring discharge that dominates Shelter Run base flow. It is possible that the relatively low concentrations and low δ³⁴S values of SO₄⁼ in spring discharge and Shelter Run base flow may reflect those of atmospheric deposition before the middle of the 20th century. In addition to storage in soils and biota, variations in ground-water residence times at a wide range of time scales may have important effects on monitoring, modeling, and predicting watershed responses to changing atmospheric deposition in small watersheds.