Optical absorption spectra of Na/Ar systems are calculated by combining the classical Monte Carlo simulation method with a quantum mechanical first-order perturbation scheme [Balling and Wright, J. Chem. Phys. 79, 2941 (1983)] for estimating the energies of the Na-* 3p(P-2) excited states. The model incorporates many drastic approximations, but contains no adjustable parameters. Our Na/Ar matrix simulations generated relaxed structures for several candidate trapping sites based on various sized vacancies in face-centered-cubic (fee) solid Ar. Trapping sites for which the equilibrium structures belong to the O-h or T-d point groups yielded the experimentally well-known "triplet" absorption line shape; for these cases, the splitting of the degeneracy of the excited Na-* 3p(P-2) state is due solely to fluctuations away from the equilibrium structures. Simulations of Na/Ar clusters, surfaces, and matrix sites possessing a strong permanent axial asymmetry yielded a widely split "doublet plus singlet" absorption line shape. Despite our success at reproducing several qualitative aspects of the absorption spectroscopy of Na/Ar matrices? our simulations failed to quantitatively reproduce the experimental data. We discuss the major limitations of our model, as well as several possible improvements.