Annual spring-seeded forage crops use less water than cereal grains, including durum (Triticum turgidum L. var. durum), and may be suitable to replace summer fallow. We conducted an experiment from 2002 through 2006 comparing yield, quality, and water and N use of durum and three annual forages, barley (Hordeum vulgare L.), barley interseeded with pea [Pisum sativum L. ssp. arvense (L.) Poir.], and foxtail millet [Setaria italica (L.) Beauv.] in 2-yr rotations. Durum in rotation with summer fallow and alfalfa (Medicago sativa L.) were included. Averaged over 5 yr, alfalfa had higher forage yield and quality, water use, and N accumulation compared to annual forages. Annual forages had similar preplant and postharvest soil water contents, but barley and barley–pea had higher yields and water use compared to millet. Barley–pea intercrop had superior forage crude protein (CP), neutral detergent fiber (NDF) and N accumulation compared to barley and millet, but acid detergent fiber (ADF) and nitrogen recovery index (NRI) were similar among annual forages. Averaged over 4 yr, preplant soil water and residual N content were greater for durum following fallow than for durum following annual forages, resulting in reduced fertilizer N requirement and greater yield, water use, grain N accumulation and NRI following fallow. Replacing summer fallow with annual forages reduced durum grain yield by 727 kg ha−1 but provided forage yield of 4.9 Mg ha−1 Annualized net returns in annual forage-durum systems were $127 ha−1, $77 and $34 ha−1 greater than for fallow-durum and alfalfa, respectively. Replacing summer fallow with annual forages reduced durum yield but improved profitability.