Slices from rat midbrain containing the raphe nuclei and from hippocampus were prepared, loaded with [3H]5-HT and superfused and the resting and the electrically stimulated [3H]5-HT release was measured. The 5-HT3 receptor agonist 2-methyl-5-HT (1 to 10 micromol/l) increased the resting tritium outflow in superfused raphe nuclei slices, EC50 5.3 micromol/l. The 2-methyl-5-HT-induced increase of tritium outflow was an external Ca2+-independent process and was not altered by reserpine pretreatment but it was reversed by addition of the 5-HT uptake inhibitor fluoxetine (1 micromol/l). The 5-HT3 receptor antagonists ondansetron and GYKI-46 903 (1 micromol/l) did not antagonize the stimulatory effect of 2-methyl-5-HT on resting tritium outflow. 2-Methyl-5-HT in lower concentration increased the electrically induced tritium overflow from raphe nuclei slices (EC50 0.56 micromol/l) and also from hippocampal slices preloaded with [3H]5-HT. These effects were reversed by 1 micromol/l of ondansetron and GYKI-46903. The 5-HT3 receptor antagonists (1 micromol/l) were without effects on depolarization-evoked [3H]5-HT release at 2 Hz stimulation, when 10 Hz stimulation was used, ondansetron and GYKI-46 903 reduced the tritium overflow from raphe nuclei slices. These data indicate that 5-HT3 receptors positively alter depolarization-induced somatodendritic 5-HT release in the raphe nuclei. They also show that 2-methyl-5-HT is able to evoke 5-HT release not only from vesicles but also from cytoplasmic stores via a transporter-dependent exchange process.