Pregnant Sprague-Dawley CD rats were administered phenytoin by gavage on days 7-18 of gestation in doses of 0, 100, 150 and 200 mg/kg. At 46-53 days of age rat offspring in each litter were divided with half tested in a Biel water maze and half in a new, more complex water maze. On the first test day all rats were tested in a straight channel. On subsequent days rats received 5 trials in the Biel maze or 6 trials in the new maze in path A, followed by 6 trials in path B (in which the start and goal positions were reversed). Analyses showed that phenytoin increased offspring mortality shortly after birth at the highest dose, but did not affect growth at any dose. Phenytoin had no effects on straight channel swimming performance. Analyses of maze performance showed that both mazes differentiated phenytoin offspring from controls, but that the new, more complex maze distinguished the groups more definitively. Maze errors also showed a clear dose-response relationship in the new maze, while no such distinctions were seen in the Biel maze. It was concluded that the new maze design may offer significant gains over the Biel maze in terms of detection sensitivity to prenatally induced CNS injury.