Even if there are evidences that the intensity of social interactions between partners vary with time, very few genetic studies have investigated how social genetic effects (SGEs) vary over time. To overcome this issue, the objectives of the present study were to analyse longitudinal records of average daily gain (ADG) in rabbits and to evaluate, by simulation, the response to selection for such longitudinal trait. Five weekly ADG records from 3,096 rabbits under feed restriction after weaning and raised in pen of 8 were used for the analysis. A linear animal mixed model including SGEs with week specific random effects that follow structured antedependence (SAD) functions was fitted to the data using ASReml and the Fortran program that we have developed (freely available on zenodo). The social heritability was higher in week 1 (0.44) than in weeks 2 to 5 (ranging from 0.16 to 0.23). The correlation between the SGEs of different weeks was moderate to high for weeks 2 to 5 (0.62 to 0.91) and weaker between the first week and the other weeks (0.33 to 0.47). The direct-social genetic correlations were negative at any time. Based on the same data design, the same variance components but considering 3 different sets of direct-social genetic antagonism (strong, moderate, weak); we simulated 7 generations of selection using a SAD model including SGEs or not to estimate breeding values. Results obtained showed that the increase in ADG with selection decreased with the direct-social genetic antagonism and was improved (~by 30%) when SGEs were taken into account. In conclusion, results confirmed that SGEs vary over time and do not correspond to the same trait after mixing than later in life, probably as a consequence of social hierarchy establishment observed at that time. Accounting for SGEs in the selection criterion maximizes the genetic progress.