In this paper, we develop an Adaptive Multigrid method (AMGM) to model photon transport through biological tissues in bioluminescence tomography. In our method, the smoothing operation on fine levels and residual correction on coarse levels in V-Cycle offer fast convergence rate for this forward problem. Using a heterogeneous phantom, the methodology is validated by Monte Carlo simulations, and the computation speed is much higher than conventional smoothing iteration methods on a single grid. In actual biomedical imaging applications, especially when there are many sources in small animal body, AMGM is potential to accurately simulate the forward problem at very low computation cost.