This paper presents a physically based method for estimating chlorophyll concentration in closed forest canopies. The radiative transfer model SAILH and the leaf optical properties model PROSPECT were coupled and inverted on MIVIS hyperspectral data acquired on July 2003 over the Ticino Park. Different inversion strategies were tested in order to predict the canopy chlorophyll concentration in oak forest sites. Estimates were validated against in situ measurements. Results show that this methodology allows to accurately retrieve chlorophyll concentration (R 2 = 0.73, RMSE = 4.0 μg/cm 2, EF = 0.48). Furthermore, our results suggest that model performances increase using prior knowledge in the inversion procedure.