Accurate calculation of internal fluence excited in tissue from an optical source can be used for predicting the performance of fluorescent contrast agents for clinical applications. Solutions of excitation fluence for a steady-state Monte Carlo model and a finite element implementation of the 3d diffusion equation have been compared up to depths of 20mm from a point source located on top of a homogeneous cylindrical phantom for a range of reduced scattering-to-absorption ratios. Differences between the fluence calculated by Monte Carlo and diffusion model is found to be dependent on the transport mean free path (mfp), size of the phantom in relation to the penetration depth, distance from the source and mesh resolution. The differences are small at depths ~ mfp and peak at depths ~2mfp. The differences should ideally reduce to zero at large depths but the magnitude of the differences tend to increase due to the finite boundary in the diffusion model. As an example, for a mfp = 0.817mm similar in magnitude to mesh resolution, diffusion fluence at 1mm, 2mm, 10mm and 14mm is 76%, 59%, 66% and 63% respectively of Monte Carlo fluence. For large mfp's characteristic of non- diffusive regimes, diffusion model overestimates fluence at distances less than one mfp. This work demonstrates that mean free path and mesh resolution are the critical parameters that distinguish the performance of Monte Carlo and diffusion models to define error margins that could be utilized for predictive assessment of imageability of fluorescent agents using the diffusion model.