Molecular imaging is an emerging imaging technique in biological and medical field. Thereinto, bioluminescence tomography (BLT) plays a significant role. In view of the ill-posedness of the BLT problem, a priori knowledge is indispensable to reconstruct bioluminescent source uniquely and quantitatively. In this paper, the anatomical information of a real mouse is obtained with the microCT scanner to represent different macroscopic biological tissues. The proposed tomographic algorithm based on the adaptive finite element methods (FEMs) employs the microCT slices based coarse volumetric mesh to reconstruct source distribution quantitatively according to a posteriori error estimation techniques. In order to avoid the inverse crime, a Monte Carlo (MC) method based virtual optical environment, molecular optical simulation environment (MOSE), is also adopted for producing the measurement data. Finally, simulation results with the above framework demonstrate the effectiveness and potential of the proposed adaptive tomographic algorithm.