Measurements of the cosmogenically-produced ³⁵S, a radioisotope of sulphur (t_1/2 = 87 days), are reported for the Ned Wilson Lake watershed in Colorado. The watershed contains two small lakes and a flowing spring presumed to be representative of local ground water. The watershed is located in the Flattops Wilderness Area and the waters in the system have low alkalinity, making them sensitive to increases in acid and sulphate deposition. Time series of ³⁵S measurements were made during the summers of 1995 and 1996 (July–September) at all three sites. The system is dominated by melting snow and an initial concentration of 16–20 mBq L^-1was estimated for snowmelt based on a series of snow samples collected in the Rocky Mountains. The two lakes had large initial ³⁵S concentrations in July, indicating that a large fraction of the lake water and sulphate was introduced by meltwater from that year's snowpack. In 1995 and 1996, ³⁵S concentrations decreased more rapidly than could be accounted for by decay, indicating that other processes were affecting ³⁵S concentrations. The most likely explanation is that exchange with sediments or the biota was removing ³⁵S from the lake and replacing it with older sulphate devoid of ³⁵S. In September of 1995 and 1996, ³⁵S concentrations increased, suggesting that atmospheric deposition is important in the sulphate flux of these lakes in late summer. Sulphur-35 concentrations in the spring water were highly variable but never higher than 3.6 mBq L^-1 and averaged 2 mBq L^-1. Using a simple mixing model, it was estimated that 75% of the spring water was derived from precipitation of previous years.