Characterizing the geothermal system at Surprise Valley (SV), northeastern California, is important for determining the sustainability of the energy resource, and mitigating hazards associated with hydrothermal eruptions that last occurred in 1951. Previous geochemical studies of the area attempted to reconcile different hot spring compositions on the western and eastern sides of the valley using scenarios of dilution, equilibration at low temperatures, surface evaporation, and differences in rock type along flow paths. These models were primarily supported using classical geothermometry methods, and generally assumed that fluids in the Lake City mud volcano area on the western side of the valley best reflect the composition of a deep geothermal fluid. In this contribution, we address controls on hot spring compositions using a different suite of geochemical tools, including optimized multicomponent geochemistry (GeoT) models, hot spring fluid major and trace element measurements, mineralogical observations, and stable isotope measurements of hot spring fluids and precipitated carbonates. We synthesize the results into a conceptual geochemical model of the Surprise Valley geothermal system, and show that high-temperature (quartz, Na/K, Na/K/Ca) classical geothermometers fail to predict maximum subsurface temperatures because fluids re-equilibrated at progressively lower temperatures during outflow, including in the Lake City area. We propose a model where hot spring fluids originate as a mixture between a deep thermal brine and modern meteoric fluids, with a seasonally variable mixing ratio. The deep brine has deuterium values at least 3 to 4‰ lighter than any known groundwater or high-elevation snow previously measured in and adjacent to SV, suggesting it was recharged during the Pleistocene when meteoric fluids had lower deuterium values. The deuterium values and compositional characteristics of the deep brine have only been identified in thermal springs and groundwater samples collected in proximity to structures that transmit thermal fluids, suggesting the brine may be thermal in nature. On the western side of the valley at the Lake City mud volcano, the deep brine-meteoric water mixture subsequently boils in the shallow subsurface, precipitates calcite, and re-equilibrates at about 130 °C. On the eastern side of the valley, meteoric fluid mixes to a greater extent with the deep brine, cools conductively without boiling, and the composition is modified as dissolved elements are sequestered by secondary minerals that form along the cooling and outflow path at temperatures <130 °C. Re-equilibration of geothermal fluids at lower temperatures during outflow explains why subsurface temperature estimates based on classical geothermometry methods are highly variable, and fail to agree with temperature estimates based on dissolved sulfate-oxygen isotopes and results of classical and multicomponent geothermometry applied to reconstructed deep well fluids. The proposed model is compatible with the idea suggested by others that thermal fluids on the western and eastern side of the valley have a common source, and supports the hypothesis that low temperature re-equilibration during west to east flow is the major control on hot spring fluid compositions, rather than dilution, evaporation, or differences in rock type.