In the dry-seeded, flooded rice (Oryza sativa, L.) production areas of the southern USA, a flood is normally maintained until shortly before harvest to minimize rice yield or milling quality reduction due to water deficits. Research conducted in Arkansas on Crowley (fine, montmorrillonitic, thermic Typic Albaqualfs) silt loam and Sharkey (very-fine, montmorillonitic, nonacid, thermic Vertic Haplaquepts) clay soils indicated no reduction in rough rice yields when flooding was ceased at heading, but effects of early cessation of flooding on rice milling quality were not assessed. The primary objective of this research was to determine if early cessation of flooding would reduce rice grain yield or milling quality. Field experiments were conducted at three Arkansas locations in 1987 and 1988. The experimental design was split-plot, randomized complete block. Main plot treatments were times of draining (TD) which were 0, 2, and 4 wk after the 50% heading date. Subplot treatments were times of harvest (TH) which were 3, 4, 5, and 6 wk after heading. Draining at 2 wk after heading did not reduce rough rice yields or head rice yields over draining at 4 wk after heading at any location in any year. On the Crowley soil, draining at heading reduced rough rice by 15 to 40 cwt/acre and head rice yields by 0 to 40% depending on year and TH. In 1987, draining at heading on the Sharkey soil reduced rough rice yields from 0 to 25 cwt/acre, and reduced head rice yields 0 to 30% depending on TH. Draining at heading decreased neither rough rice yields nor head rice yields on the Calhoun soil or in 1988 on the Sharkey soil. It appears that rice growers producing rice in the Mid-South on these or similar soils may safely cease flooding 2 wk after heading.