On the basis of preliminary results concerning the analysis of lucerne variability for quality traits, it seems worthwhile paying more attention to the variability between entries than within: within entry variability seems quite difficult to handle, even if the absence of a negative correlation of statistical significance between dry matter yield and crude protein content deserve a certain interest. To this respect, a large number of single plants should be analysed to apply selection differentials within lucerne populations and to start with a breeding program. Such a breeding approach could be quite expensive and it is difficult to propose to private breeders in the absence of a clear economic return in terms of higher market prices for higher quality lucerne products.