Caffeine degradation performance via photo-Fenton treatment was investigated under different dosage conditions. Experiments were planned according to a design of experiments to characterize hydrogen peroxide dosage protocols. Fenton reagent loads were first determined after a preliminary study. The addition of a fixed hydrogen peroxide load was controlled by an initial load fraction and the span of the continuous flow producing the total load. The experiments were carried out using 12 L solution samples of commercial coffee (300 mg L-1 approximately 17 mg L-1 of caffeine). Hence, HPLC data revealed that all treatments completely removed caffeine from samples, while TOC monitoring revealed reductions of up to 70%. Using reaction conversion as a performance index, the methodological approach presented enabled a comparison of dosage protocols and determination of those producing enhanced results. For the particular case addressed, the operation scheme that most increased treatment performance produced a conversion which was 25% higher than that obtained without dosage.