The quality control of product is a major and permanent preoccupation of productive enterprise in particular an agribusiness industry. Indeed, the products non-conform to the requirements for quality imposed by the standards and norms, present risks for the producers and the consumers.

Among these products and which is constitutes an important nutritional interest, the milk. In fact, its composition and its physicochemical properties, make of it a medium very favorable to the multiplication of the pathogenic micro-organisms. The quality of this product thus depends, amongst other factors of the heat treatment aiming at destroying the pathogenic bacteria lasting the phase of pasteurization.

The establishment of the thresholds values of the thermo-resistance variation parameters of  the  Pathogenic micro organisms is a paramount stage in the reduction analysis of 

contamination and toxicity risks.However, the values of these parameters are defined in 

a semi-quantitative way as intervals. Consequently, it is often difficult to stabilize exactly the values of these parameters in order to obtain a good and perfect pasteurization.

To overcome this difficulty, this article proposes the application of fuzzy logic like alternative for determining the toxicity thresholds of product (milk). The fuzzy model using fuzzy rule-based system in order to generate the parameters values intervening in

 pasteurization. The fuzzy model suggested will be checked and validated by applying it to a process of pasteurization on the Aurès-Batna dairy.

Keywords— Thresholds of toxicity, Pasteurization, Fuzzy rule-based system.