ABSTRACT Understanding the extent to which land‐use changes (LUC) impact soil organic carbon (SOC) is essential for accurate carbon accounting and global efforts aimed at reducing the negative impact of LUC on climate change. Recognizing that most of the SOC change due to LUC occurs in the topsoil, current efforts to quantify SOC change often overlook subsoils beyond 30 cm depth. We used data from Germany's national agricultural and forest soil inventories to address this sparsity by modeling depth‐dependent, LUC‐induced, SOC stock change down to 90 cm using data‐driven reciprocal modeling. This modeling was carried out using an ensemble approach for prediction and area of applicability assessments to avoid extrapolation. Landscape, climate, and pedological properties were used to predict the equilibrium SOC stock at four depth intervals (0–10, 10–30, 30–60, and 60–90 cm) for all six land‐use change directions between cropland, grassland, and forest. While the greatest change occurred at the surface for all LUC directions, we detected significant SOC stock change down to the sampled depth of 90 cm. Approximately 30% of the detected SOC stock change was found in the subsoil (30–90 cm). For LUC to or from forests, the litter layer dominated the changes in SOC such that for LUC between grassland and forest, SOC stock change in the mineral soil was mostly offset by the addition or removal of the litter layer. For all LUC directions, the World Reference Base soil group was the most important factor for determining the magnitude of SOC stock change. This study underscores the importance of deeper soil sampling for accurate carbon accounting and climate‐change mitigation strategies.