© Yuan et al. Manipulative experiments and observations along environmental gradients, the two most common approaches to evaluate the impacts of climate change on nutrient cycling, are generally assumed to produce similar results, but this assumption has rarely been tested. We did so by conducting a meta-analysis and found that soil nutrients responded differentially to drivers of climate change depending on the approach considered. Soil carbon, nitrogen, and phosphorus concentrations generally decreased with water addition in manipulative experiments but increased with annual precipitation along environmental gradients. Different patterns were also observed between warming experiments and temperature gradients. Our findings provide evidence of inconsistent results and suggest that manipulative experiments may be better predictors of the causal impacts of short-term (months to years) climate change on soil nutrients but environmental gradients may provide better information for long-term correlations (centuries to millennia) between these nutrients and climatic features. Ecosystem models should consequently incorporate both experimental and observational data to properly assess the impacts of climate change on nutrient cycling.