The evaluation of the risk indicates a procedure based on the analysis of the risk. It is a question of considering the risks in order to treat on a hierarchical basis them and of comparing them with a level considered to be acceptable. This evaluation amount dimensioning each identified dangerous situation, at the same time compared to the gravity of its consequences and its frequency of occurrence. That supposes that it is necessary to define scales of quotation of the risk in term of frequency and gravity as well as a grid of criticality allowing the combination of these two parameters and clarifying the criteria of acceptability retained for the evaluation of the risk. However, knowledge of which we lay out concerning the probability of occurrence of the events and the gravity of their consequences is generally imperfect. The validity of the results of the evaluation depends on the taking into account total or partial of the imperfection of knowledge used. This work falls under a prospect for improvement of the conventional grids of criticality for a better control of the industrial risks. The objective being to show the interest of the fuzzy set theory in the assessment of the criticality of the risks in the presence of uncertain data.