Drought occurs often in the West African Sahel, but studies have shown that soil water availability is not usually the limiting factor to pearl millet [Pennisetum glaucum (L.) R. Br.] production, and that field water-use efficiency (WUE)-i.e., the ratio of yield to evapotranspiration (ET)-is almost always very lo,v. The purpose of this study mas to determine management effects on yield and water use of pearl millet for a range of climate conditions in the Sahel. Grain and aboveground dry matter yield, daily vapor pressure deficit, and soil mater data were taken during four gears of contrasting rainfall. Within any given year, genotype, plant population, and fertilizer had relatively small to no effect on ET, but large effects on yield. When high plant population (greater than or equal to 20 000 hill ha(-1)) was combined with high fertilizer application (greater than or equal to 40 kg N ha(-1) and greater than or equal to 18 kg P ha(-1)) during the wettest year, total ET was increased by approximate to 50 mm. High fertilizer application tended to slightly increase ET and thereby deplete soil water reserves, but this was not associated with yield decline. Yield and water-use data refute the view that, by maintaining fields at low fertility and low plant populations, farmers reduce risk of crop failure during drought by reducing crop water use. Compared with traditional practices that use plant populations as low as 5000 hill ha(-1) and zero fertilizer input, moderate plant population (10 000 hill ha(-1)) and fertilizer application (20 kg N ha(-1) and 9 kg P ha(-1)) substantially increased yield and approximately tripled WUE even during 1984, the driest year on record. In general, grain yield was better predicted from ET within different management categories when corrections were made for mean daily vapor pressure deficit during the growing season ((VPD) over bar). The study provides evidence for the need to moderately increase pearl millet plant population and fertilizer application in the Sahel to efficiently use available water without risk of crop failure through depletion of soil water reserves. It also provides a practical, albeit empirical, basis for predicting yield under different management systems from seasonal ET and (VPD) over bar data.