Bomb radiocarbon, 14C, produced by atmospheric weapons testing in the early 1960’s has been long recognized as an excellent tracer for the study of exchanges between the carbon reservoirs and the global carbon cycle on decadal time-scales. SOM exchanging with the atmosphere reflect the rate of exchange through the amount of ‘bomb’ 14C incorporated. The 14C content of a homogeneous C reservoir in any given year since 1963 may be predicted from the turnover time and the known record of atmospheric 14CO2. 14C as a continuous isotopic label can be used as an extremely useful tool to determine the dynamics of soil carbon and validate model structure and parameters. Observed ecosystem carbon fluxes and stocks data were used to calibrate and validate the PnET-SOM model to accurately simulate ecosystem carbon dynamics for the Harvard Forest deciduous forest. Observed 14C signature data in different soil organic matter pools (1996 and 2007) were then used to compare the predicted 14C by the model to validate the parameter values and model structure. The model showed reasonably the temporal dynamics of the 14C label as it moved from the surface litter and roots into the mineral soil organic matter pools. The model correctly simulated the impact of bomb 14C on the soil organic matter pools in the forest floor, which directly interact with fresh litters and have faster turnover rates. However, the model underestimated the 14C signatures of SOM pools in mineral soils, especially the light fractions. It implies that either the SOM turnover were too slow, the sources of the carbon had low 14C signature, or both. The current model structure of mineral associated SOM could not both accurately predict carbon stocks and 14C signatures. The dissolved organic carbon from recent carbon pools out of the litter layer into the mineral soil, and mixing of the humus layer into the mineral soil layer could be the solution to reconcile the observed carbon fluxes, stocks, and 14C signature. More data analyses and modelling practices are expected in the future to improve the model.