Surface water draining granitic bedrock in Yosemite National Park exhibits considerable variability in chemical composition, despite the relative homogeneity of bedrock chemistry. Other geological factors, including the jointing and distribution of glacial till, appear to exert strong controls on water composition.  Chemical data from three surface water surveys in the upper Merced River basin conducted in August 1981, June 1988 and August 1991 were analysed and compared with mapped geological, hydrological and topographic features to identify the solute sources and processes that control water chemistry within the basin during baseflow.  Water at most of the sampling sites was dilute, with alkalinities ranging from 26 to 77 μequiv. l⁻¹. Alkalinity was much higher in two subcatchments, however, ranging from 51 to 302 μequiv. l⁻¹. Base cations and silica were also significantly higher in these two catchments than in the rest of the watershed. Concentrations of weathering products in surface water were correlated to the fraction of each subcatchment underlain by surficial material, which is mostly glacial till. Silicate mineral weathering is the dominant control on concentrations of alkalinity, silica and base cations, and ratios of these constituents in surface water reflect the composition of local bedrock.  Chloride concentrations in surface water samples varied widely, ranging from <1 to 96 μequiv. l⁻¹. The annual volume-weighted mean chloride concentration in the Merced River at the Happy Isles gauge from 1968 to 1990 was 26 μequiv. l⁻¹, which was five times higher than in atmospheric deposition (4–5 μequiv. l⁻¹), suggesting that a source of chloride exists within the watershed. Saline groundwater springs, whose locations are probably controlled by vertical jointing in the bedrock, are the most likely source of the chloride.  Sulphate concentrations varied much less than most other solutes, ranging from 3 to 14 μequiv. l⁻¹. Concentrations of sulphate in quarterly samples collected at the watershed outlet also showed relatively little variation, suggesting that sulphate may be regulated to some extent by a within-watershed process, such as sulphate adsorption.