Hydrogen (H<sub>2</sub>) is under consideration as a green energy source for future low-carbon development pathways. However, assessment of the future climate impacts of large-scale H<sub>2</sub> use is currently limited by poor understanding of the dominant atmospheric removal process, the uptake of H<sub>2</sub> by microbial activity in soil. Here we implement a soil dry deposition scheme for H<sub>2</sub> in a global chemistry transport model and conduct a sensitivity analysis to quantify how uncertainties in soil parameters contribute to uncertainty in H<sub>2</sub> uptake. We find that soil moisture dominates the uncertainty (52 %), along with contributions from soil porosity (34 %) and the water threshold for biological activity (13 %). However, the sensitivity of uptake to soil moisture is strongly nonlinear, with increases in uptake with declining soil moisture where it is under biotic control and with increasing soil moisture where it is under diffusivity control. We show that soil uptake dominates the interannual variation in atmospheric H<sub>2</sub> abundance from 2010 to 2022, and the notable decrease in 2015, while enhanced atmospheric production drives the observed long-term increase. Our results suggest that both soil sink and source of H<sub>2 </sub>are affected by strong ENSO events driven by changes in soil moisture. Using output from 11 CMIP6 models, we project changes in H<sub>2</sub> soil uptake between 2015 and 2100 of −2.5 % to +7.9 % under SSP1-2.6 and +4.1 % to +20.4 % under SSP5-8.5, suggesting an increase in soil uptake under climate change associated with future changes in soil moisture.