Wistar rats were exposed to a multiple fixed-time 30-s food delivery schedule, with an on/off tone signalling the two components. Animals were matched in accordance with the levels of schedule-induced polydipsia. Drinking was then punished in one of the components: half of the rats received lick-dependent 10-s signalled delays and the other half lick-dependent electric shocks. The intensities of the shocks were adjusted to reduce behaviour by the same amount as the delays in food presentation. Unpunished components were used as yoked-control conditions, by presenting delays or shocks independently of the animals' behaviour. D-Amphetamine (0.3–2.0 mg/kg) and cocaine (1.0–10.0 mg/kg) dose-dependently increased (although only slightly) and then decreased schedule-induced polydipsia punished with lick-dependent delays in food presentation, a result not observed in control conditions or when the behaviour was suppressed by lick-dependent electric shocks. Diazepam (1.0–17.0 mg/kg) and pentobarbital (3.0–17.0 mg/kg) dose-dependently increased and then decreased only the schedule-induced drinking punished with lick-dependent shocks. Buspirone (0.1–1.0 mg/kg) and morphine (2.0–5.6 mg/kg) showed either no specific effects or further suppressed schedule-induced drinking. Results of these and previous experiments suggest that the antipunishment effects of drugs depend not only on the precise nature of the drug, but also on the manner in which the behaviour is maintained.