In current clinical practice, photodynamic therapy (PDT) is carried out with prescribed drug doses and light doses as well as fixed drug–light intervals and illumination fluence rates. This approach can result in undesirable treatment outcomes of either overtreatment or undertreatment because of biological variations between different lesions and patients. In this study, we explore the possibility of improving PDT dosimetry by monitoring drug photobleaching and photoproduct formation. The study involved 60 mice receiving the same drug dose of a novel verteporfin-like photosensitizer, QLT0074, at 0.3 mg/kg body weight, followed by different light doses of 5, 10, 20, 30, 40 or 50 J/cm² at 686 nm and a fluence rate of 70 mW/cm². Photobleaching and photoproduct formation were measured simultaneously, using fluorescence spectroscopy. A ratio technique for data processing was introduced to reliably detect the photoproduct formed by PDT on mouse skin in vivo. The study showed that the QLT0074 photoproduct is stable and can be reliably quantified. Three new parameters, photoproduct score (PPS), photobleaching score (PBS) and percentage photobleaching score (PBS%), were introduced and tested together with the conventional dosimetry parameter, light dose, for performance on predicting PDT-induced outcome, skin necrosis. The statistical analysis of experimental results was performed with an ordinal logistic regression model. We demonstrated that both PPS and PBS improved the prediction of skin necrosis dramatically compared to light dose. PPS was identified as the best single parameter for predicting the PDT outcome.