Concentrations of dissolved organic carbon (DOC) in surface waters are increasing across Europe and parts of North America. Several mechanisms have been proposed to explain these increases including reductions in acid deposition, change in frequency of winter storms and changes in temperature and precipitation patterns. We used two modelling approaches to identify the mechanisms responsible for changing surface water DOC concentrations. Empirical regression analysis and INCA-C, a process-based model of stream-water DOC, were used to simulate long-term (1986--2003) patterns in stream water DOC concentrations in a small boreal stream. Both modelling approaches successfully simulated seasonal and inter-annual patterns in DOC concentration. In both models, seasonal patterns of DOC concentration were controlled by hydrology and inter-annual patterns were explained by climatic variation. There was a non-linear relationship between warmer summer temperatures and INCA-C predicted DOC. Only the empirical model was able to satisfactorily simulate the observed long-term increase in DOC. The observed long-term trends in DOC are likely to be driven by in-soil processes controlled by SO4(2-) and Cl(-) deposition, and to a lesser extent by temperature-controlled processes. Given the projected changes in climate and deposition, future modelling and experimental research should focus on the possible effects of soil temperature and moisture on organic carbon production, sorption and desorption rates, and chemical controls on organic matter solubility.