Studies rarely assess treatment effects across tasks; the present experiments addressed this issue. In Experiments 1 and 2, rats (n=12) were trained and then tested with variable delays on a spatial match-to-position task sequentially in the water and radial mazes (in counterbalanced order). Experiment 1 compared the effect of 0-, 60- and 1440-min delays on performance in both mazes. Rats required fewer (P<0.05) mean (+/-S.E.M.) sessions to reach criterion performance in the water (11.0+/-1.0) versus radial maze (19.3+/-2.2). In test sessions, performance was impaired delay-dependently when scores were averaged across the two tasks (P<0.05) but no significant effect of task or task x delay interaction was found. In the second experiment, the same rats were retrained and tested with 0-, 1-, 3- and 5-min delays in both mazes and testing followed the administration of scopolamine (0, 0.1, 0.4 and 0.8 mg/kg). The mean (+/-S.E.M.) number of acquisition sessions was similar in the radial (6.33+/-0.34) and water maze (6.08+/-0.46). On the sample portion of trials, performance was impaired at the 0.8 mg/kg dose of scopolamine (P<0.02) in the radial maze only. On the recognition portion of trials in the radial maze, the 0.4 and 0.8 mg/kg doses of scopolamine impaired performance whereas in the water maze task the 0.8 mg/kg dose impaired performance. The pattern of results may reflect different natural tendencies of rats to alternate (win-shift) versus not alternate (win-stay) in dry land versus swim tasks.