Forest soils store large amounts of carbon (C), and releases of C from this pool may significantly increase the CO 2 concentration in the atmosphere. Organic matter decomposition in soils has been shown to strongly depend on temperature and soil moisture and is, therefore, susceptible to the climate change. Reliable methods are needed to monitor and predict the changes in soil C stocks. In this study, we tested the Yasso07 soil C model by comparing the model predictions to repeated soil C measurements of organic layer and, furthermore, to the estimates of two other C models, namely Yasso and ROMUL. In the model simulations, we used the litter input time series derived from forest biomass estimates based on the national forest inventories. Both the repeated empirical measurements and Yasso07 simulations indicated upland forest soils to be small sinks of C in Southern Finland. The Yasso07 model was able to predict both soil C stock and C accumulation within the error limits of the measured values. Yasso07 and the earlier ver-sion, Yasso, predicted very similar soil C stocks close to the measured values, but slightly underestimated C accumulation. The annual soil C changes predicted by the Yasso07 and ROMUL models were reasonably close to each other, even though the models are based on a very different basic structure. However, the dif-ferences in the model predictions were at the highest in years with the highest precipitation, indicating that there are still uncertainties in predicting the effects of soil moisture on the soil C stock changes. Ó 2012 Elsevier B.V. All rights reserved.