We present the first quantitative study of polymorphic content in a model pharmaceutical formulation using transmission Raman spectroscopy (TRS), and compare the results obtained with those from traditional backscattering geometry. The transmission method is shown to provide a true bulk measurement of the composition, being unaffected by systematic or stochastic sub-sampling issues that can plague traditional backscattering geometries. The accuracy of the quantification of the polymorphs using TRS was shown to surpass considerably that achieved using conventional backscattering mode. For a model-free fit, the TRS method yielded R(2) of 0.996 compared to the backscattering value of 0.802; for a partial least squares fit with a single component the TRS method accounted for 98.09% of the variance in the data and yielded an R(2) of 0.985, compared to 89.65% of the variance and R(2) of 0.804 for the backscattering method.