1. The objective of the experiments was to explore the modulatory functions of the serotonergic cerebral giant cells (CGCs) of the Lymnaea feeding system by examining their synaptic connections with the central pattern generator (CPG) interneurons and the modulatory slow oscillator (SO) interneuron. 2. One type of modulatory function, 'gating,' requires that the CGCs fire tonically at a minimum of 7 spikes/min. Above this minimum level the CGCs control the frequency of CPG interneuron oscillation-'frequency control,' a second type of modulation. In an SO-driven fictive feeding rhythm, an increase in the frequency of the rhythm, with increased CGC firing rate, resulted from a reduction in the duration of the N1 (protraction) and N2 (rasp) phases of the feeding cycle with little effect on the N3 (swallow) phase. 3. The CGCs excited the N1 phase interneurons SO and N1M (N1 medial) cells but had no consistent effects on the N1 lateral cells. The CGC → SO postsynaptic response was probably monosynaptic (≤200 ms in duration) with unitary 1:1 excitatory postsynaptic potentials (EPSPs) following each CGC spike. The CGC → N1M excitatory response was slow and nonunitary, and a burst of CGC spikes evoked a depolarization of the N1M cells that lasted up to 10 s and triggered N1M cell bursts. Both CGC → SO and CGC → N1M excitatory responses could be mimicked by the focal application of serotonin (5-HT). 4. Both CGC → SO and CGC → N1M excitatory connections systematically increased the N1M cell firing rate within the CGCs' physiological firing range (0-40 spikes/min). This was due to both the direct (CGC → N1M) and indirect (CGC → SO → N1M) excitatory synaptic pathways. The CGC-induced increase in N1M cell firing rate probably accounted for the reduced duration of the N1M cell feeding burst by causing a more rapid reversal of the feeding cycle from the N1 phase to the N2 phase. This phase reversal was due to the previously described recurrent inhibitory pathway (N1 → N2 excitation followed by N2 → N1 inhibition). 5. The CGCs' ability to provide a depolarizing drive to the N1M cells meant that this excitatory connection was also likely to be important for gating. 6. Activity in the CGCs produced nonunitary, long-lasting, excitatory postsynaptic responses on the N2 ventral (N2v) CPG interneurons, and these were likely to be involved in both the gating and the frequency control by the CGCs on the N2 phase of the feeding rhythm. Suppressing CGC tonic firing initially increased the duration of the N2v plateau (which determines the duration of the N2 phase of the feeding cycle, frequency function) but eventually led to a loss of N2v plateauing (gating function). 7. Nonunitary, weakly inhibitory CGC → N2 dorsal responses were recorded that could be mimicked by the application of 5-HT. 8. Spikes in the CGCs evoked 1:1 monosynaptic EPSPs in the N3 tonic (N3t) CPG interneurons. This excitatory effect could be mimicked by the application of 5-HT. Within the physiological range of CGC firing, this excitation did not appear to influence the firing rate of the N3t cells. 9. N3 phasic (N3p) CPG interneurons showed biphasic (hyperpolarizing followed by depolarizing) unitary responses to spikes evoked in the CGCs. The inhibitory synaptic response was maintained in a high-Ca2+/high-Mg2+ (Hi-Di) saline and was mimicked by the focal application of 5-HT, indicating that it was probably monosynaptic. The excitatory component was, however, reduced in a Hi-Di saline, indicating that it was probably polysynaptic. Suppressing the CGCs during an SO-driven feeding rhythm caused the N3p cells to fire less, suggesting that the removal of the excitatory component of the response might be significant. 10. We conclude that the general depolarizing effects of the CGCs on a number of the CPG cell types may provide one explanation of the CGCs' ability to gate the feeding CPG. These excitatory responses fell into two classes. Either the recorded responses were unitary, lasting ~200 ms (cf. SO and N3p and N3t cells), or the responses lasted much longer, ~- 10 s (cf. N1M and N2v cell types). Gating may also involve affects on the endogenous properties of a subset of the CPG interneurons. These include the N1M cells (triggering bursts) and N2v cells (necessary for plateauing). Several of the CGCs' effects on the CPG interneurons could explain their ability to control the frequency of an SO-driven rhythm. First, by exciting the SO, N1M, and N2v cell types, they could control the duration of the N1 phase of the feeding cycle. Second, by influencing the duration of the N2v plateau they could determine the duration of the N2 phase of the feeding cycle. 11. CGC stimulation or the application of 5-HT causes responses on the CPG interneurons that last a maximum of 10 s. However, the effects seen after suppression of CGC activity take many minutes to have complete effect.