Serotonin and 5-HT3 receptors may be involved in the activation of nociceptive afferent pathways by rectal distension. In rats, intracolonic infusion of glycerol is able to trigger nociceptive inputs as evidenced by the occurrence of abdominal constrictions. This work was designed to evaluate the influence of 5-HT3 receptor antagonists on this reflex and to approach the site of action by comparing their relative efficacies according to the route of administration. Male Wistar rats (250-350 g) were surgically prepared for abdominal electromyography and a catheter was placed in the colonic lumen. Five days after surgery, electrical activity of abdominal muscles was recorded before and during (20 min) intracolonic infusion of glycerol (60% glycerol + 40% saline, rate 0.75 mL/h). Cilansetron was administered intraperitoneally, 15 min before glycerol infusion, at doses of 5 to 500 micrograms/kg. Granisetron, ondansetron and cilansetron were administered at the dose of 20 micrograms/kg by intraperitoneal (i.p.), intravenous (i.v.) or intracolonic (i.c.) routes. The number of abdominal spike bursts was used as an index of visceral nociception. Intracolonic infusion of glycerol increased significantly (P < 0.05) the number of abdominal spike bursts during the time of infusion compared with saline (30.6 +/- 6.6 vs 4.5 +/- 3.4 bursts). When administered i.p., cilansetron dose-dependently reduced the frequency of abdominal spike bursts from the dose of 20 micrograms/kg i.p. Administration i.p. of granisetron and ondansetron at this dose also significantly reduced the number of abdominal spikes (19.0 +/- 6.0 and 18.3 +/- 6.9 respectively). Cilansetron, ondansetron and granisetron were also effective by i.v. and i.c. routes, cilansetron was more active by the i.c. route. Serotonin, via 5-HT3 receptors, is involved in the mediation of abdominal contractions induced by intracolonic infusion of glycerol. 5-HT3 receptor antagonists are also active by i.c. route suggesting a local site of action.