The performance of three pure-bred and five crossbred groups of ewes was evaluated under three different production systems, i.e., intensive, semi-intensive and extensive. The breed groups were the Cyprus Fat-tailed (L), Chios (C), Awassi (A), first generation crosses (C x L, A x L and A x C) and two backcrosses to Chios (C x C x L and C x A x C). Only Chios and Awassi ewes were represented in the intensive production system. All three pure-breds and crossbred groups were represented in the extensive system, whereas the two backcrosses were not present in the semi-intensive system. Feeding was based on concentrates, straw and hay (cereal and alfalfa) and quantity and quality was regulated according to the stage of the production cycle within a production system. Production systems significantly affected litter weight at birth and at weaning (P less than 0.01) and milk production (P less than 0.01). There were no serious effects on litter size either at birth or at weaning (P greater than 0.05). Chios ewes had a higher litter size and litter weight at birth and at weaning than both other pure-breds and most crossbred groups. The milk production of Chios and Awassi ewes was similar and higher than that of the Local and A x L breed groups. The most serious interaction effects between genetic groups and production systems occurred in milk production and litter weight at weaning. There was a change in rank between two crossbred genotypes (C x L and A x L) and a change in scale involving the Chios and Awassi pure-breds. The question, however, is how to overcome the problems of genotype x environment interactions. Should we breed for a particular environment or improve the environment when better genotypes are introduced to unfavourable ones.