Organic photochemistry becomes a key synthesis pathway for sustainable chemistry thanks to its ability to address molecular complexity and diversity with a “flick of a switch”, often in one step. Sensitized photooxygenations are particularly attractive: singlet oxygen 1O2 is here generated by photosensitization of the triplet molecular oxygen 3O2, most often in the visible range with catalytic amounts of an organic sensitizer (dye). Despite the several advantages, these reactions have not found widespread implementations in the chemical industry, mainly due to currently available technology based on outdated batch reactors (poor light penetration, high dilution) equipped with energy demanding mercury lamps (intensive cooling and optical filter needs). In that way, continuous flow microstructured technologies have recently emerged as alternatives to batch processing and their suitability for photoreactions has been highlighted; the combination with narrowly emitting LED light sources additionally enables energy savings, increased yields and selectivity. At present, however, there are few attempts to understand these benefits using a chemical engineering perspective. In keeping with this context, this thesis investigates photooxygenation in different continuous-flow microreactor devices with the aim at (i) understand how and why the operating conditions can affect the yield and the selectivity; and (ii) providemodelling tools for defining a methodology for smart scale-up. For that, three types of LED-driven microreactors were considered: the Vapourtec® photoreactor, the home-made spiral-shaped photoreactor and the advanced-flow G1 photoreactor from Corning®. As a case of study, the photooxygenation of -terpinene into ascaridole was selected, in the sustainable solvent ethanol and using rose Bengal as a sensitizer of industrial interest. In a first step, a revised protocol of actinometry using the Reinecke’s salt (visible domain) was proposed to measure the incident photon flux (i. e the one arriving at the reactor surface). It was successfully implemented in the spiral-shaped microreactor, and based on simplified 1D model, this parameter could be estimated. Secondly, the kinetic law based on all mechanistic steps was established for the chosen reaction. From this, the set of operating conditions potentially influencing the photoreaction rate were identified. Subsequently, experiments with soluble Rose Bengal were carried in both lab-scale and meso-scale microreactors, enabling to cover a large operating window and thus to collect an important database. Whatever the continuous-flow reactors, the conditions enabling minimization of sensitizer bleaching while achieving quantitative conversion were identified. A 1D model of representation, taking into account the Lambertian nature source and the potential reflection at the back optical surface, was proposed. It enabled to formalize the different coupled phenomena, to analyze and estimate the experimental trends at different scales. Finally, the photooxygenation of -terpinene was performed under continuous-flow with solid-supported sensitizer. The latter consisted in photoactive microgels (~300 nm) transported by the liquid phase: the sensitizer Bengal rose was grafted on a polymer matrix that had the ability to swell into water or alcoholic solvents and thus induced a decrease of their average refraction index. The underlying idea was to test a new-resource efficient photochemical synthesis concept that could circumvent the downstream separation processes and also protect the sensitizer from its ability to photodegrade during the irradiation time. The implementation of these photoactive microgels lead to very promising results, in terms of both conversion, bleaching reduction and reusability, and thus offer very interesting perspectives for supported photooxygenation in the future.