Soil carbon (C) sequestration has been proposed as a transitional win-win strategy to help replenish organic-matter content in depleted agricultural soils and counter increases in atmospheric greenhouse gases. Data assimilation and remote sensing can reduce uncertainty in sequestered C mass estimates, but simple soil organic carbon (SOC) models are required to make operational predictions of tradeable amounts over large, heterogenous areas. Our study compared the performance of RothC26.3 and a reduced compartmental model on an 11-year fertilizer trial in subhumid West Africa. Root mean square error (RMSE) differences of 0.05 Mg C/ha between models on total SOC predictions suggest that for contractual purposes, SOC dynamics can be simulated by a two-pool structure with labile and stable components. Faster (seasonal) and slower (semicentennial and beyond) rates can be approximated by constants as instantaneous and infinite decay. In these systems, simulations indicate that cereal residue incorporation holds most potential for mitigation of transient C loss associated with recent land conversion to agriculture.