The controlling effects of soil moisture on soil nitrogen (N) supply to plants have been well established from lab experiments, and are often incorporated in soil organic matter (SOM) models, but rarely tested in the field along natural gradients of soil moisture. With a large dataset of in-situ incubation experiments across a wide range of natural ecosystems (mostly with shallow groundwater) in the Netherlands and Belgium, we tested whether and how soil moisture influences N mineralization rates (Nmin) in natural soils, and what type of hydrological coupling (in terms of estimation of soil moisture and formulation of soil moisture effect) is suited to predict across-site variation of Nmin with a SOM model, CENTURY. Our results show that soil moisture was related to a small (17%), but a larger than other site factors, proportion of the difference between measured and reference (i.e. simulated with no moisture effects) Nmin, indicating that soil moisture has a subtle yet important influence on Nmin of natural soils. Coupling CENTURY with a simple tipping-bucket hydrological module, a common way to estimate soil moisture in existing models, did not improve model performance on Nmin compared to a model without soil moisture effects, irrespective of the formulation of moisture effect applied. Coupling of CENTURY model with a more detailed hydrological module improved the model performance, but only when an appropriate moisture function was applied. Our results also indicated that model performance largely improved if indirect moisture effects through N losses via denitrification were accounted for. With our best model, inclusion of soil moisture information improved the normalized residual variance from 28% to 18% of the observed range of Nmin. We conclude that appropriate estimates of soil moisture and proper representation of its effects on mineralization are crucial to properly reflect between-site variation in soil N supply in natural ecosystems.