Abstract: The assessor performances is a key point in a sensory evaluation. In particular, at the end of a session, a decrease of the performance can be feared. We propose to analyse this performance with various criteria: usual ones as the main product effect or the error variance; a new one measuring the perceived products variability. The performance can be then studied all along the session from two points of view: in taking into account the only products tested at a given instant (named instantaneous); in taking into account all the products tested up to a given instant (named cumulative). In the presented example, in spite of the large number of products successively tested by each assessor, the instantaneous performance of the panel show no significant deterioration. Furthermore, when the number of products tested by each assessor increases, more significant product effects can be obtained thanks to the accumulation of the amount of data. This shows that the number of products that can be reasonably studied by one assessor during one session is generally underestimated.