Mercury capture from coal-fired power plant flue gas in the ductwork and on the fabric filter by powdered activated carbon injection was simulated by means of a detailed adsorption model. The model is based on material balances in both gaseous and adsorbed phases along the duct/ filter length and inside the activated carbon particles. The growing filter cake moving boundary problem was solved with a double orthogonal collocation technique after a suitable immobilization of the moving front. Model results indicated that high mercury removal efficiencies in the duct are only obtained with the use of large sorbent loadings, because of the short gas/sorbent contact time. On the contrary, effective gas/sorbent contact in the fabric filter leads to high removal efficiencies with moderate sorbent consumption. In both cases, the sorbent feed rate can be lowered by selecting a reactive sorbent and by decreasing the sorbent average particle size or the operating temperature. Model results for in-duct mercury capture are validated against a bench scale experimental set of data recently reported in the literature. Further comparison of model predictions with available pilot- and full-scale data suggests that simulation of mercury capture in real power plants will require taking into account the additional mercury removal by deposited activated carbon on the duct walls.