Photodynamic Inactivation (PDI) is a promising alternative technique to conventional methods for the control of microorganisms in food. The present work proposed a mathematical model that can predict the photo inhibitory activity of the combination of eosin Y and green LED light against Staphylococcus aureus ATCC 25923. A Rotational Central Composite Design was used to establish the conditions for the photo inhibitory activity of eosin Y against S. aureus. Twelve experiments, including 3 replicates of the central point, were conducted to evaluate the effects of two independent variables: the concentrations of dye (1.60 to 4.98 µmol L-1) and the lighting time (4.3 to 15.6 min). Results were expressed as Log CFU/mL. Reductions in bacterial counts were observed even at low concentrations of dyes and lighting times, proving the efficacy of PDI mediated by eosin Y against S. aureus. The mathematical model proposed was significant (p=0.000034) and didn’t show lack of fit (p=0.4014). These models were able to explain up to 91.59% of the data variance. Although the adjusted coefficients of determination (R2) of the mathematical models were 0.9159, considering the significance of prediction, lack of fit of this model and the natural variability related to the microbiological experimentation, the model was considered adequate to predict the photo inhibitory activity. The variable dye concentration showed linear and quadratic negative effects, however the linear effect is more significant (p= 0.0023 and 0.0295, respectively). The negative linear effect was also observed for the illumination time (p = 0.0062). This means that as the dye concentration, or the illumination time, increases there is a tendency to decrease the bacterial count. The interaction effect of the two variables was not significant (p>0.05). The mathematical models can predict the values of concentration of PS and the time of illumination necessary to reach certain bacterial inhibition.