Wood duck (Aix sponsa) broods were counted 15 of 19 years (1973-90) using the night-float technique on a 1O.8-km segment of the Holston River, eastern Tennessee. Broods were counted 9 years before and 6 years after the initiation of September hunting seasons. Mean harvest (¯x = 26,797, SE = 2,130) of wood ducks in Tennessee after September hunting began was 279% higher (t = 7.50, P = 0.0001, df = 14) than the mean harvest (¯x = 9,604, SE = 845) before September hunting. The adjusted mean number of broods (¯x = 7.25, SE = 1.25) observed following initiation of September hunting was significantly (F (1,8) = 16.78, P = 0.00035) less (52%) than the adjusted mean (¯x = 15.22, SE = 1.49) observed prior to September seasons. These results suggest a negative relationship between numbers of broods observed and September hunting.