Field studies were conducted in 1984 and 1985 using small plots (non-weighing lysimeters) of 1 m 2 to determine the effect of four levels (L) and three times (T) of soil moisture stress treatments on yield and quality of Maple Amber soybean [ Glycine max (L.) Merr.]. The study incorporated the most probable combinations of these moisture stress treatments. Thousand-seed weight was inversely related to L ( n 2 = 0.98). The L × T effect on seed yield was significant ( P = 0.01). Seed yield decreased continuously from 1863 to 847 kg/ha (mean of 2 years) with increasing moisture stress levels when irrigation was applied at each prescribed L, regardless of growth stage ( r 2 = 0.93). In this case, however, irrigation produced no appreciable yield advantage above that obtained at 33% moisture stress level, or greater, when moisture stress at growth stage R2 was induced. In the latter case, a linear relationship of yield with moisture stress ( r 2 = 0.96) was obtained. Constant but low yields were maintained across the moisture stress range by withholding irrigation at R5, even though adequate water was applied to plants up to the induced stress. Protein was mainly affected by L ( P = 0.01). A curvilinear relationship of protein with moisture stress ( r 2 = 0.99) demonstrated that the highest seed protein content is associated with a well-watered condition maintained by frequent irrigation. The oil, in contrast, was affected by T ( P = 0.01). The highest yields together with the highest seed protein could be achieved under well-watered conditions. Under a high water deficit, in contrast, only moderate protein content was obtained, but yield was reduced significantly.