This work describes the critical role of the matching between the spectral distribution of the incident light and the absorption spectra of the semiconductor material on the accuracy of the simulation of solar photocatalytic reactors. In silico results have been generated by a multiphysics model including the rigorous description of the hydrodynamics, radiation transfer, mass transport and chemical reaction rate, with the particular feature of using a spectral band discretization approach for the resolution of the radiation balance and the estimation of the spectral local volumetric rate of photon absorption. Model predictions with a mechanistic kinetic model have been experimentally validated using a solar reactor based on a tubular reactor coupled to a compound parabolic collector (CPC) under illumination with a solar simulator (xenon lamp) and natural sunlight. Small differences in the spectral distribution of both light sources led to significantly different predictions for the reaction rate, explaining the higher efficiency experimentally achieved with natural sunlight in comparison with the solar simulator for equivalent total UV-A irradiances. The developed model is able to explain the discrepancies between experimental results in solar simulators and under natural sunlight, usually reported in the literature, offering a novel numerical approach for rigorous modelling of photoreactors using light sources with any spectral distribution.