Previous research has shown that systemic injections of the nonselective serotonin (5-HT) antagonist metergoline (MET) can stimulate feeding in both rats and humans. Five experiments were conducted to determine if this drug would elicit feeding in otherwise satiated rats after direct injections into the brain. In experiment 1, intraventricular infusions of 100 nmol MET produced reliable enhancements of feeding for 1 h compared with control infusions of a 5% tartaric acid (vehicle) solution. In experiment 2, a dose-response study of 0, 50, 100, and 150 nmol MET intraventricularly revealed that both 100 and 150 nmol doses reliably enhanced 1-h feeding, whereas 50 nmol did not. In experiment 3, tests of 90-min locomotor activity and water intake in the absence of food revealed that 100 nmol MET intraventricularly did not modify either behavior compared with vehicle infusions, suggesting a degree of feeding specificity to this effect. In an attempt to determine the site of intraventricular MET effects on feeding, experiment 4 tested 1-h feeding responses after 0, 5, 10, 20, 40, or 60 nmol MET were infused unilaterally into the paraventricular nucleus (PVN) of the hypothalamus. No reliable feeding was induced at any of these drug doses, although injections of 30 nmol norepinephrine (NE) were effective in doubling food intake. Experiment 5 further showed that bilateral infusions of 50 nmol MET in each PVN (total dose, 100 nmol) were ineffective in altering 1-h feeding. This contrasted markedly to the high potency of 15 nmol NE per site (total dose, 30 nmol), which elicited fivefold greater feeding than control infusions.(ABSTRACT TRUNCATED AT 250 WORDS)