Manifold advantages are foreseen by using carbon nanotubes (CNTs) as support for inorganic TiO2 nanoparticles due to the unique texture/morphology and adsorption capacity of CNTs. Synergistic effects might also result from interfacial charge transfer between the CNTs and TiO2. Effective charge transfer has the potentiality to limit electron/hole recombination and shift the TiO2 photocatalytic response to the visible range. Homogeneous mixing and intimate contact between the graphitic and TiO2 surfaces are of high importance in order to trigger synergistic effects. Thus, the existence of complementary methods to shed light on both these features is of high importance when developing TiO2/CNT composite photocatalysts. In this work, a wide variety of TiO2/CNT composites was prepared by a simple hydration/dehydration procedure, using single-wall (SWCNTs) and multi-wall (MWCNTs) carbon nanotubes, either functionalized or not, and TiO2 nanoparticles of different size. To evaluate the degree of homogeneity between the graphitic and inorganic phases, a new methodology which was based on a complex interpretation of the liquid nitrogen porosimetry (LN2) isotherms of the composites and of each phase in the composite separately was developed. Furthermore, interface interaction characteristics were elucidated by micro-Raman spectroscopy while small-angle X-ray scattering (SAXS) measurements provided insight on the surface roughness and micropore structure of the TiO2/SWCNT samples. The Raman analysis concluded to the absence of any interfacial interaction. In this context the efficiency of the prepared composites to photocatalytically oxidize caffeine was evaluated in regard to their homogeneity, as derived by the LN2 method. As expected, in the absence of synergetic effects the photocatalytic efficiency correlated well with the extent of mixing between the CNTs and TiO2 phases. The discrepancy observed for one of the samples was attributed to the existence of large micropores, a feature that was distinguishable solely by SAXS measurements.