Inland waters are key components of the global carbon cycle, emitting a substantial amount of carbon dioxide (CO2) to the atmosphere. Climate change and global warming affect mountain catchments significantly, leading to glacier retreat and changing precipitation patterns. What this means for the streams in the mountains, and for CO2 sources, dynamics and fluxes from those streams, is poorly understood. Generally, mountain streams have low CO2 concentrations, largely due to low organic carbon content in catchments above the tree line and poorly developed soil horizons. However, it is yet unclear from where the streamwater CO2 is coming, what drives the CO2 dynamics, and the magnitudes of the CO2 evasion fluxes from the stream surfaces to the atmosphere. This PhD thesis aims to fill this knowledge gap. By using a combination of field sampling, sensor monitoring stations, and analyses of stable isotopes, I identify potential drivers of CO2 concentrations and fluxes from mountain streams across different spatial and temporal scales. Moreover, in combination with recent insights of gas exchange from mountain streams, this thesis comprises new quantifications of CO2 evasion fluxes from Swiss mountain streams, as well as mountain streams worldwide. Within the frame of this thesis, two major field studies were carried out. The first field study consisted of repeated seasonal sampling campaigns every 50 meter across an Alpine stream network. We found large seasonal and spatial variations in streamwater CO2 concentrations, and identified soil respiration as the major source of streamwater CO2. With spring snowmelt, and increased catchment connectivity, more respiratory CO2 could be transported to the stream, however due to high gas transfer velocities, the CO2 was rapidly evaded downstream. The second major field study consisted of high-frequency (10 minutes) monitoring of streamwater CO2 in 12 streams located in 4 Alpine catchments. This enabled us to further explore the role of snowmelt for CO2 dynamics in mountain streams. We identified different responses in CO2 concentration and CO2 evasion fluxes to increasing runoff across different temporal scales. On annual time scales, increasing runoff led to lower streamwater CO2 concentration. However, during the onset of snowmelt we found increasing CO2 concentrations with higher runoff, leading to higher CO2 evasion fluxes. To better contextualize our results and estimate CO2 evasion fluxes from mountain streams, we combined insights gained from the two field studies with recent insights of gas exchange in turbulent streams. We developed a simple CO2 prediction model, and modelled the CO2 evasion fluxes from all small mountain streams in Switzerland, as well as all mountain streams worldwide. We estimate that mountain streams worldwide emit 167 ± 1.5 Tg C yr-1. This estimate is unexpectedly high given the small total surface area of small mountain streams. This thesis brings new light on the CO2 dynamics in mountain streams, not only by highlighting the role of spatiotemporal catchment dynamics and in particular the important role of snowmelt for facilitating transport of catchment-derived CO2 to the streams, but also by providing one of the first global estimates of the quantity of CO2 that is evaded from mountain streams every year.