The development of swimming behavior and the correlated activity patterns recorded in motoneurons during fictive swimming in paralyzed zebrafish larvae were examined and compared. Larvae were studied from when they hatch (after 2 days) and are first capable of locomotion to when they are active swimmers capable of capturing prey (after 4 days). High-speed (500 Hz) video imaging was used to make a basic behavioral characterization of swimming. At hatching and up to day 3, the larvae swam infrequently and in an undirected fashion. They displayed sustained bursts of contractions ('burst swimming') at an average frequency of 60-70 Hz that lasted from several seconds to a minute in duration. By day 4 the swimming had matured to a more frequent and less erratic "beat-and-glide" mode, with slower (approximately 35 Hz) beats of contractions for approximately 200 ms alternating with glides that were twice as long, lasting from just a few cycles to several minutes overall. In whole cell current-clamp recordings, motoneurons displayed similar excitatory synaptic activity and firing patterns, corresponding to either fictive burst swimming (day 2-3) or beat-and-glide swimming (day 4). The resting potentials were similar at all stages (about -70 mV) and the motoneurons were depolarized (to about -40 mV) with generally non-overshooting action potentials during fictive swimming. The frequency of sustained inputs during fictive burst swimming and of repetitive inputs during fictive beat-and glide swimming corresponded to the behavioral contraction patterns. Fictive swimming activity patterns were eliminated by application of glutamate antagonists (kynurenic acid or 6-cyano-7-nitroquinoxalene-2,3-dione and DL-2-amino-5-phosphonovaleric acid) and were modified but maintained in the presence of the glycinergic antagonist strychnine. The corresponding synaptic currents underlying the synaptic drive to motoneurons during fictive swimming could be isolated under voltage clamp and consisted of cationic [glutamatergic postsynaptic currents (PSCs)] and anionic inputs (glycinergic PSCs). Either sustained or interrupted patterns of PSCs were observed during fictive burst or beat-and-glide swimming, respectively. During beat-and-glide swimming, a tonic inward current and rhythmic glutamatergic PSCs (approximately 35 Hz) were observed. In contrast, bursts of glycinergic PSCs occurred at a higher frequency, resulting in a more tonic pattern with little evidence for synchronized activity. We conclude that a rhythmic glutamatergic synaptic drive underlies swimming and that a tonic, shunting glycinergic input acts to more closely match the membrane time constant to the fast synaptic drive.