Different management practices were evaluated in a 2-year field experiment for their effect on the water use of pearl millet grown in farmers' fields in two village sites (Banizoumbou and Bagoua) of South western Niger. These practices were laid out in a Randomized Complete Block Design comprising 5 manure treatments (blocks), no manure, 1-year corralling, 2-year corralling, 3-year corralling, and transported manure; 3 fertilizer treatments, no fertilizer, 2 g di-ammonium phosphate, and 2 g di-ammonium phosphate + 1 g urea; and finally 2 millet varieties, a local variety and an improved variety, Zatib. Each experimental plot was 10 m x 10 m. Profile soil moisture content was monitored weekly up to a depth of 2.1 m, from the beginning to the end of the season, using neutron scattering. Rainfall was measured at each of the sites using a rain gauge. During the second season, destructive leaf area index measurements were done fortnightly. Water balance calculations were done to estimate evapotranspiration and deep drainage losses. Results of soil profile moisture remaining at the end of the season were analyzed using SPSS. Manure treatment had the most influence on the amount of soil profile moisture remaining at the end of the season (p < 0.05) for both seasons during the two seasons, followed by variety treatment. In both years and at both sites, the no manure treatment, which is typical of farmers' management had the most amount of residual profile moisture and incidentally the lowest yields. There was a significant difference between the two sites (p < 0.05) in both millet grain yields and water use. Addition of manure and using improved millet varieties is a good method to improve yields and water use in Sahelian agriculture.