Dryland ecosystems are experiencing more variability and extremes in rainfall and disproportionate shifts in plant community composition, both likely to alter soil carbon (C) cycling and storage. Despite these trends, we lack long-term experimental data that facilitates predicting shifts in ecosystem function with climate change. This dataset records changes in soil organic carbon (SOC) and inorganic carbon (SIC) storage in the top 1 m of soil profiles following 19 years of experimental manipulation of rainfall and vegetation within a cold-desert ecosystem. A split plot design was employed (n = 3) and included 1) contrasting vegetation types (split plots), either native Artemisia tridentata spp. tridentata (big sagebrush) communities or monocultures of Agropyron cristatum (crested wheatgrass), a non-native bunchgrass, and 2) manipulations of spring/fall or summer rainfall (whole plots). We further stratified the plots by under-plant vs. inter-plant patches. Soil C responses to long-term rainfall treatments varied by vegetation type. Long-term summer rainfall treatments significantly increased both SOC and SIC pools under A. tridentata, with total carbon (TC) pools 1.15 × ambient controls (P = 0.02). Carbon pools in spring/fall rainfall treatments significantly decreased, with TC pools 0.80 × ambient (P = 0.05) due to losses of inorganic carbon. In contrast, A. cristatum increased in SOC but lost SIC in response to both summer and spring/fall rainfall additions, resulting in no change to a slight gain in TC pools (P = 0.29). Both SOC and SIC pools in inter-plant spaces increased with summer rainfall treatments and decreased with spring/fall rainfall regardless of vegetation type. In contrast to most studies that only examine surface soils (0-0.1 m), our findings indicate that increases in cool-season rainfall will cause A. tridentata communities to become a net C source, whereas A. cristatum monocultures may become C sinks – largely due to tradeoffs between SOC and SIC pools. We conclude that consideration of vegetation type, the entire vertical profile, and both organic and inorganic C forms are imperative to predicting responses of dryland ecosystems to changing climate.