I investigated controls of stream dissolved inorganic carbon (DIC) sources and cycling along a stream size and productivity gradient in a temperate forested watershed in northern California. Dissolved CO2 (CO2 (aq)) dynamics in heavily shaded streams contrasted strongly with those of larger, open canopied sites. In streams with canopy cover > 97%, CO2 (aq) was highest during baseflow periods (up to 540 ??M) and was negatively related to discharge. Effects of algal photosynthesis on CO2 (aq) were minimal and stream CO2 (aq) was primarily controlled by groundwater CO2 (aq) inputs and degassing losses to the atmosphere. In contrast to the small streams. CO2 (aq) in larger, open-canopied streams was often below atmospheric levels at midday during baseflow and was positively related to discharge. Here, stream CO2 (aq) was strongly influenced by the balance between autotrophic and heterotrophic processes. Dynamics of HCO3- were less complex. HCO3- and Ca2+ were positively correlated, negatively related to discharge, and showed no pattern with stream size. Stable carbon isotope ratios of DIC (i.e. ??13C DIC) increased with stream size and discharge, indicating contrasting sources of DIC to streams and rivers. During summer baseflows, ??13C DIC were 13C-depleted in the smallest streams (minimum of -17.7???) due to the influence of CO2 (aq) derived from microbial respiration and HCO3- derived from carbonate weathering. ??13C DIC were higher (up to -6.6???) in the larger streams and rivers due to invasion of atmospheric CO2 enhanced by algal CO2 (aq) uptake. While small streams were influenced by groundwater inputs, patterns in CO2 (aq) and evidence from stable isotopes demonstrate the strong influence of stream metabolism and CO2 exchange with the atmosphere on stream and river carbon cycles.