In this paper the suitability of three perennial, herbaceous, lignocellulosic grasses (Arundo donax, Saccharum spontaneous spp. aegyptiacum and Miscanthus x giganteus) for the production of second-generation bioethanol in semi-arid Mediterranean environment was studied. Crops were established in spring 2002, supplying irrigation and nitrogen fertilization up to 2004/2005 growing season. Subsequently, crops were grown without any agronomic input and harvested annually. Data reported in this paper refers to 2008/2009 and 2009/2010 growing seasons. Aboveground dry matter (DM) yield was higher in Arundo (35.4±2.1 Mg ha-1 in 2009 and 32.2±1.9 Mg ha-1 in 2010 harvest) than in Saccharum (27.3±2.0 and 23.9±1.9 Mg ha-1, respectively) and Miscanthus (19.6±2.8 and 17.2±1.6 Mg ha-1, respectively). Structural polysaccharides of the raw material were higher in Miscanthus (63.4 %w/w) followed by Saccharum (61.5% w/w) and Arundo (57.6% w/w). The same trend was identified for the cellulose content (41.0%, 36.8% and 34.6%, respectively). The highest values in the total hemicellulose complex were observed in Saccharum (24.7%), followed by Arundo (23.1%) and Miscanthus (22.4%). The composition of structural polysaccharides leads to a higher theoretical ethanol yield (TEY) from one dry ton of Miscanthus feedstock (kg DM Mg-1), followed by Saccharum and Arundo. On the other hand, the TEY per unit surface (Mg ha-1) was greater in Arundo than in Saccharum and Miscanthus. When compared to other lignocellulosic sources used in the second-generation bioethanol technology, such as agricultural residues, woody species and other herbaceous perennial crops, Arundo, Saccharum and Miscanthus showed a great potential in terms of TEY ha-1. Given the high levels of biomass yield and composition of structural polysaccharides, the three species might be introduced into the Mediterranean cropping systems to supply lignocellulosic biomass for second-generation industrial plants or bio-refineries.