This study aims to test the influence of size grading on self-feeding behaviour, social structure (measured by the percentage of triggering acts per individual), growth performances, and blood physiological variables of individually passive integrated transponder (PIT)-tagged sea bass juveniles, using a computerized on-demand feeding system coupled with a PIT tag monitoring device. Three consecutive periods of 27 days each were compared: a first period (P1) before grading (6 tanks of 100 fish; 40.2 +/- 8.9 g) followed by a second period (P2) after grading. The protocol applied aimed to create two groups of fish of similar mean weight but with either a low or a high coefficient of variation of weight (CV(w)) corresponding to an imposed difference in social disruption (T(low): CV(w) similar to 10%, 3 tanks of 60 fish each with social disruption; T(high): CV(w) similar to 20%, 3 tanks of 60 fish each, without social disruption). T(low) and T(high) groups were studied over P2, and an additional 27-day period under identical conditions (P3). The grading protocol used and/or time modified the social structure when comparing P1 and P2. Thereafter, during P2 and P3, no difference could be observed in growth performances, feed demand, or physiological variables between T(low) and T(high) groups. Feeding rhythms and social structures were similar in both groups. In conclusion, such grading practice only transiently modifies feed demand behaviour and social structure built around the self-feeder, without further improvement in individual growth performances in sea bass.