In this article, a complete radiative transfer approach for estimating incident photon flux density by actinometry is presented that opens the door to investigation of large-scale intensified photoreactors. The approach is based on an original concept: the analysis of the probability that a photon entering the reaction volume is absorbed by the actinometer. Whereas this probability is assumed to be equal to one in classical actinometry, this assumption can no longer be satisfied in many practical situations in which optical thicknesses are low. Here we remove this restriction by using most recent advances in the field of radiative transfer Monte Carlo, in order to rigorously evaluate the instantaneous absorption probability as a function of conversion. Implementation is performed in EDStar, an open-source development environment that enables straightforward simulation of reactors with any geometry (directly provided by their CAD-file), with the very same Monte Carlo algorithm. Experimental investigations are focused on Reinecke salt photodissociation in two reactors designed for the study of natural and artificial photosynthesis. The first reactor investigated serves as reference configuration: its simple torus geometry allows to compare flux densities measured with quantum sensors and actinometry. Validations and analysis are carried out on this reactor. Then, the approach is implemented on a 25 L photobioreactor with complex geometry corresponding to one thousand light-diffusing optical fibers distributing incident photons within the reaction volume. Results show that classical actinometry neglecting radiative transfer can lead to 50 percent error when measuring incident flux density for such reactors. Finally, we show how this radiative transfer approach paves the way for analyzing high conversion as a mean to investigate angular distribution of incident photons. (C) 2016 Elsevier B.V. All rights reserved.