A split-plot experiment was conducted in thermally controlled chambers using Columbian Plymouth Rock chickens to determine the effect of water-cooled roosts on performance in hot ambient conditions. The birds were subjected to 25 +/- 1, 35 +/- 1, and 25 +/- 1 C ambient temperature treatments for 2, 3, and 2 wk, respectively. Roost temperature treatments were either cool (20 +/- 1 C) or air-equilibrated (25 +/- 1 or 35 +/- 1 C). The performance parameters evaluated were percentage hen-day egg production, egg weight, feed intake, feed conversion, fertility, and hatchability. Birds subjected to the water-cooled roost treatment had consistently higher performance than birds using the air-equilibrated roost under all three ambient temperatures. Both ambient and roost temperature treatments significantly influenced percentage hen-day egg production, average daily feed intake, and percentage hatchability (P less than or equal to .05). However, the biggest differences in performance were observed during the heat-stress period. Decreases in performance during the heat-stress period from the thermoneutral control values were: 5.95 and 13.1 percentage points for hen-day egg production, 22.2 and 34.8 percentage points for average daily feed intake, and 5.17 and 15.38 percentage points for hatchability in water-cooled and air-equilibrated roost treatments, respectively. The ambient and roost temperature treatments did not significantly affect egg weight, feed conversion, or percentage fertility. The improved performance with water-cooled over air-equilibrated roost treatments, especially during heat-stress periods, indicates that the water-cooled roosts minimized the deleterious effects of heat stress through conductive heat loss from the birds to the roost.