In order to improve the agricultural water use efficiency, a multi-objective optimization model for optimizing agricultural water resources allocation was established considering the whole process of allocation of agricultural water resources. The research objects include eight canals and two grain crops (spring wheat and mulched maize) under the main canal of Huangyang Irrigated Area in Liangzhou District, Wuwei City, Gansu Province. The objective function is to minimize the irrigation loss of main and branch canals and to maximize the economic benefit of main grain crops. We set three kinds of different initial soil moisture content when solving the multi-objective model using geometric weighting method and minimum deviation method respectively. The results showed that the minimum deviation method performed better, and the weight of the objective function should not be biased towards minimizing the loss of irrigation, otherwise the economic benefits would decrease. Under the same ETa, the pre-sowing irrigation saved 286.08×104 m3 of water than the post-sowing irrigation, so the water-saving effect of pre-sowing irrigation was better. On the premise of ensuring regional food security, the planting structure should be adjusted to planting more maize and less wheat to obtain more economic benefits. The allocation scheme of agricultural water resources proposed can not only increase the yield of grain crops and farmers′ profit, but also provide theoretical basis and guidance for the decision making of this irrigated area management.