Random regression models are widely used to analyze longitudinal data in genetic evaluation. They permit to evaluate more precisely the environmental effects and the genetic value of an animal, by taking into account the test-day specific effects. Our objective was to implement a random regression model for the evaluation of French Saanen goats. The data consisted of milk test-days (TD) records from first lactation of Saanen goats. B-splines were used to model the environmental factors. The genetic and permanent environment effects were modeled through two Legendre polynomials of the same order. The goodness of fit and the genetic parameters being function of the order of the polynomials, orders 0, 2 and 4 were tested. Models with Legendre polynomials of order 4 were found to be the most parsimonious. A rank reduction of the variance covariance matrix was performed by eigenvalues decomposition in order to reduce computing time and complexity. With a reduction to rank 2, the first two eigenvectors well summed up the genetic variability of milk yield level and persistency.