The degree of alternation of arm choice in a Y-maze was examined on 15-min tests over 4 days in rats treated (IP) with saline, amphetamine (0.5 or 2.0 mg/kg) or pretreated with haloperidol (0.08 mg/kg) in each condition prior to test. On day 1 amphetamine-treated animals chose arms at random, but from day 2-4 those receiving the higher dose perseverated their choice. Controls maintained alternation. These effects could be prevented by haloperidol pretreatment. Amphetamine treatment increased the frequency of rearing at the middle, choice-point of the maze more than at the end of an arm. The increase at the mid-point was suppressed by haloperidol pretreatment from day 1 and at the end of an arm from day 2. Amphetamine induced an increase in head-turning/"looking" that was suppressed by haloperidol from day 2. The effect of haloperidol in increasing the duration of an item of looking or rearing at the end of an arm also started later in testing. Two effects are postulated to have occurred: (i) a conflict on day 1 between novelty-controlled sensory or attentional effects that leads to an alternation of arm choice and amphetamine-induced dopaminergic activity that facilitates an alternation of behavioural responses. The result was random choice and increased rearing at the choice point. (ii) On days 2-4 the drug-induced effects on switching motor responses came to control behaviour.