Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) has been established as a good and relatively sensitive method for qualitative analysis of powders and rough surfaces. The technique is non-destructive and high quality spectra can be easily obtained without extensive sample preparation. The capacity of the method for qualitative analysis of functional groups has been demonstrated in numerous fields. The potential of DRIFTS for quantitative analysis was recognized quite early as well. However, many theoretical and practical problems do appear when the method is applied for this purpose, requiring very accurate sampling procedure. The effect on reflectance spectra of several parameters influencing sampling has been studied. Conventionally, a simplified relation derived from the phenomenological continuum theory of Kubelka-Munk is used in order to relate the concentration of a chromophore, dispersed in a sample, to the intensity of the diffusely reflected radiation. In practice however, this equation can be used only for ideal systems diluted in a non-absorbing scattering matrix and in a limited concentration range. Preparation of such idealized samples in a reproducible way is difficult. Quantitative analysis of opaque non-diluted systems such as those found in the real world is of high interest. In many fields, like for example in thin layer chromatography, stationary phase studies, environmental pollutant analysis, and industrial quality control or heterogeneous catalysis, it is desirable to quantitatively characterize samples as close as possible to their original state. In our own effort to do quantitative spectroscopy of powders, we have investigated the case of non-diluted samples. Model systems constituted of two types of silica powders (Cab-O-Sil and LiChrosorb) of very different morphology were prepared. The surface hydroxyl groups of SiO2 were exposed to a silylation reaction with an aminosilane containing a suitable chromophore. Known superficial concentrations of molecules carrying a cyano functional group were anchored by this way on the powder surface. A quantitative model of diffuse reflectance, derived from the general hyperbolic solutions of the Kubelka-Munk equations has been developed, taking into account different parameters such as the absorption of the silica substrate, finite thickness of the sample, specular reflection, as well as diffusion and absorption coefficients profiles within the depth of the scattering layer. Application of this model gives a quite good description of measured diffuse reflectance, in cases where the simple Kubelka-Munk relation fails. Use of a Mar quardt type non-linear least square adjustment method to the experimental data allows moreover the determination of other parameters of interest. A new FTIR analytical technique, based on a PVDF photo-pyroelectric film detector, has also been developed, which has enabled for the first time sequential measurement of diffuse reflectance and diffuse transmittance of the same sample without any change in the optical geometry and sample manipulation. Another multilayer discontinuum theory, more suited for particles of diameter larger than the radiation wavelength, has been developed and successfully applied to diffuse reflectance and transmittance measurements of sodium carbonate powders.