The Songwe geothermal prospect is situated in western Tanzania in the Rukwa Rift of the western branch of the East African Rift System. Thermal springs discharge along NW–SE oriented fracture zones in two separate areas: in the main Songwe graben (Iyola, Main springs, Rambo and Kaguri) and eastern Songwe graben (Ikumbi). Lithologies forming and filling the Songwe graben are metamorphic gneiss and shist as basement rocks, overlain by the Karoo sandstones, and Red sandstones, both silt- and sandstones with a carbonatic matrix. In some areas of the graben, volcanic rocks intruded these formations forming basalt outflows. The discharge temperatures of springs are between 37 and 85 °C with Na-HCO 3 type fluids. Carbonate deposits surround most of the springs. Using previous geophysical, geological studies and historical fluid geochemical data and mineral data, the Songwe geothermal system interpretation was updated, including new reservoir fluid temperature, fluid flow pathway and water–rock interaction models. The classical geothermometers of K-Mg and Na-K-Ca (Mg correction) were used to predict the reservoir fluid temperature and show that fluid emerging in the Songwe area reaches temperatures between 125 and 148 °C. Reservoir fluid characteristics are reconstructed based on the geothermometer calculation and a PHREEQC model in which the deep fluid reacts with certain lithologies. Minerals precipitating at the surface and reservoir depth were used to calibrate the models. The models run at surface temperature were calibrated with minerals precipitating around the springs and suggest that Songwe thermal fluids interact with Red sandstone only, while Ikumbi spring water is the only spring that interacts with all lithologies (simplified referred to as: metamorphic rocks, Karoo and Red sandstone). The model run at reservoir temperature indicates that rising water is also in contact with Karoo sandstones and Ikumbi spring water composition is again influenced by the contact with all lithologies in the graben. Our conceptual model summarizes all data showing the meteoric origin of the fluids, the travel through the basement, rising along the Mbeya fault and the main reaction with sandstones through a lateral travel towards the hot springs. The proposed models reinforce the idea that carbonate dissolution from the sandstone layers is the most common water–rock interaction. Our model is supported by carbonate deposition observed in all springs, dominated by HCO 3 and Na.