An experiment was conducted in order to investigate hay yield and nitrogen harvest in binary smooth bromegrass (Bromus inermis Leyss cv. Tohum Islah) mixtures with alfalfa (Medicago sativa L. cv. Kayseri) and red clover (Trifolium pratense L. cv. Tohum Islah) in Erzurum, Turkey for 5 years between 1991 and 1995. The Hay yield, nitrogen harvest, protein concentration and land equivalent ratio (LER) in the mixtures with alternating rows of 1:1, 2:1 and 1:2 of smooth bromegrass with alfalfa and red clover were compared to those in pure legume stands without any N-fertilizer application or pure smooth bromegrass stands that received 0, 50, 100 and 150 kg ha(-1) N. The mixtures had no N fertilization apart from 40 kg N ha(-1) in the establishment year. The dry matter production in all the mixtures receiving no N fertilizer application was higher than in pure legume stands. Pure grass stands were sustained only with the application of 150 kg ha-l N. The highest hay yields were obtained from the mixtures of smooth bromegrass (Sb) with red clover (Rc) (2Rc 1Sb) (14.65 t ha(-1)) and with alfalfa (A) (1A 1 Sb) (14.49 t ha(-1)). Although N application increased Sb yields in pure stands, the highest yields obtained with N fertilization were still lower than the yields in the mixtures without N application. The superiority of the mixtures was also reflected by their large N harvests (e.g. 355.9 kg N ha(-1) in 2Rc 1Sb plots) compared to pure Rc (317.8 kg N ha(-1)), pure A (294.3 kg N ha(-1)) and pure Sb stands that received 150 kg N ha(-1). The nitrogen harvest increased in pure Sb plots as the N doses applied increased. Furthermore, the protein concentration of the hay from the mixtures (158.2-165.7 mg g(-1)) was equal to that of the purl A stands (165.7 mg g(-1)) and higher than that of pure Sb stands (122.9 mg g(-1) at 150 kg N ha(-1) application) although the hay from pure Re plots had the highest protein concentration (179.3 mg g(-1)). The LER values were also higher in the mixtures (e.g. 1.28 in 1A 1Sb and 1.28 in 2Rc 1Sb plots) compared with the pure stands. The mixture plots also had a more balanced temporal distribution of hay. The grass component was more productive in early spring, whereas the legume fractions grew better in the summer. In conclusion, for a sustainable production of high-quality hay and greater N harvests without using N fertilizers, binary mixtures of Sb with A in alternating rows (1A 1Sb) were recommended for long-purpose stands and in alternation with double red clover rows (2Rc 1Sb) for short purpose stands under similar conditions. N application could be eliminated in the grass-legume mixtures without any yield depression. (C) 1999 Elsevier Science B.V. All rights reserved.