This study was conducted this way : one local landrace (a six-row type) and two improved cultivars (a two-row and six-row type) were either, cracked, rolled or left whole to make up nine possible dietary combinations in a 3 by 3 factorial design. Barley grain was fed along with sunflower cake, a commercial concentrate and a mineral vitamin pre-mix. Data did not show a significant difference in animal performance and carcass merits due to barley genotype and processing. Regarding in vivo DM digestibility of the ration, processing had a significant effect only on ACSAD 60 (a two-row improved cultivar) and the local landrace. In ACSAD 60, cracking gave the highest DM digestibility of the diet. With the local landrace, both processing types outperformed the whole grain. Diet digestibility with Laanaceur (a six-row improved cultivar) was not affected by processing. The results indicate that the barley breeding programme conducted at INRA for higher producing varieties does not necessarily compromise grain feeding value for animals. We can also infer from the results that grain processing, with the associated energy cost, is not necessary prior to sheep feeding.