Escalating arsenic contamination throughout Chinese soils

China faces widespread soil arsenic pollution caused by intensified industrial and agricultural activities, the impacts of which, however, have never been evaluated at the national scale. In this study, we developed a machine-learning model built on 3,524 surveys, representing over one million soil samples, to generate annual maps of arsenic concentration in China’s surface soils for the period 2000–2040. The model has uncovered a worrying trend of increasing arsenic concentrations, rising from a mean of 11.9 mg kg−1 in 2000 to 12.6 mg kg−1 in 2020, with an anticipated further increase to 13.6 mg kg−1 by 2040. The primary anthropogenic causes have been identified as non-ferrous mining activities (68.0%), followed by energy consumption (15.8%), smelting (13.2%) and farming practices (3.0%). Furthermore, in 2000, 2020 and 2040, the model predicts that 13.0%, 17.1% and 18.3% of rice production and 10.0%, 13.9% and 15.9% of the population, respectively, would be located on soils with arsenic concentrations over 20 mg kg−1. Despite the establishment of initiatives such as the Soil Pollution Prevention and Control Action Plan by the Chinese government to restrain this burgeoning arsenic pollution, our findings underscore the urgent need for more vigorous measures to stall or reverse this disturbing trend.