Redstem is an important weed in California water-seeded rice fields because of its aquatic habit, wide distribution, interference with harvest, and resistance to the herbicide bensulfuron. Our objective was to understand the mechanisms of competition for light between rice and redstem, with the goal of improving redstem control. A replicated greenhouse experiment was done in 1993 and 1994. Rice was water-seeded at a rate of 400 seeds m−2, and redstem was seeded simultaneously at approximate densities of 0, 50, and 100 seeds m−2 in continuously flooded 0.77 m2 basins. Plants were harvested once at final harvest in 1993 and twice in 1994, with an additional nondestructive sampling 34 days after seeding (DAS). Despite slower early growth, redstem height exceeded rice height about 45 DAS. At the midseason harvest in 1994 (56 DAS), no effects of redstem competition on any rice response variables were detected. However, at final harvest (110 and 118 DAS, 1993 and 1994, respectively) redstem competition at both treatment densities reduced rice tiller density, panicle density, shoot drymass, and grain drymass. Redstem competition reduced rice growth only after penetrating the canopy. Shade cast by redstem through rice maturity decreased shoot and grain production and increased tiller mortality. Lodging caused by redstem further affected rice growth. Season-long competition from redstem at mean densities of 67 and 110 plants m−2 reduced rough rice yields 31 and 39%, respectively, making redstem the most competitive broadleaved rice weed yet studied. Improved understanding of rice-redstem interactions indicates that using alternative herbicides to bensulfuron is unlikely to increase yield losses to redstem, and that control may be improved by increasing rice plant densities or slightly delaying early season chemical control. Because these strategies are mechanistic, they may also be useful for controlling other rice weeds with growth patterns similar to redstem.