Due to its unmatched hardness and chemical inertia, diamond offers many advantages over other materials for extreme conditions and routine analysis by attenuated total reflection (ATR) infrared spectroscopy. Its low refractive index can offer up to a 6-fold absorbance increase compared to germanium. Unfortunately, it also results for strong bands in spectral distortions compared to transmission experiments. The aim of this paper is to present a methodological approach to determine quantitatively the degree of the spectral distortions in ATR spectra. This approach requires the determination of the optical constants (refractive index and extinction coefficient) of the investigated sample. As a typical example, the optical constants of the fibroin protein of the silk worm Bombyx mori have been determined from the polarized ATR spectra obtained using both diamond and germanium internal reflection elements. The positions found for the amide I band by germanium and diamond ATR are respectively 6 and 17 cm(-1) lower than the true value dtermined from the k(nu) spectrum, which is calculated to be 1659 cm(-1). To determine quantitatively the effect of relevant parameters such as the film thickness and the protein concentration, various spectral simulations have also been performed. The use of a thinner film probed by light polarized in the plane of incidence and diluting the protein sample can help in obtaining ATR spectra that are closer to their transmittance counterparts. To extend this study to any system, the ATR distortion amplitude has been evaluated using spectral simulations performed for bands of various intensities and widths. From these simulations, a simple empirical relationship has been found to estimate the band shift from the experimental band height and width that could be of practical use for ATR users. This paper shows that the determination of optical constants provides an efficient way to recover the true spectrum shape and band frequencies of distorted ATR spectra.