This study concerns functional recovery of zebrafish following spinal cord transection. Spinal cords were transected at the level of the 14th vertebra, just rostral to the dorsal fin. Recovery was tested at one month after transection when descending fibers start to regrow across the transection site and at three months after transection when fish perform kick and glide swimming. To estimate the rate of regrowth across the lesion site we analysed the tyrosine hydroxylase (TH) and dorsal 5-hydroxytryptamine (5-HT) systems in distal parts of lesioned cords. Both systems have cell bodies in the brainstem and in control fish TH- and dorsal 5-HT-containing fibers descend to all spinal segments. Swimming performance was studied by subjecting lesioned fish to endurance tests in a swimming tunnel with water flowing at a constant rate of 2 or 4.5 body lengths per second (BL/s). At 2 BL/s slow myotomal muscles are active whereas at 4.5 BL/s fast myotomal muscles are recruited. Control fish endured sustained swimming at both speeds for at least 3 hours. As a measure for the condition of the neuromuscular system in trunk and tail, we analysed aerobic metabolic capacities, assessed by NADH-tetrazolium reductase (NADH-TR) histochemistry of myotomal muscle fibers and spinal lateral neuropil. We found that TH- and dorsal 5-HT-immunoreactive fibers were absent in the entire distal part of lesioned cords at one month but at two months after transection they were present at approximately 6000 microns caudally to the site of the lesion. Thus the rate of outgrowth of these fibers is at least 200 microns per day. Sustained swimming at the slow speed (2 BL/s) could be endured for about 14.4 min at one month and for 23.5 min at two months after transection; there was no further improvement in the period that followed. In contrast, in the 10 weeks following transection, fast swimming (4.5 BL/s) could be endured for about 5 to 6 minutes. A significant improvement was gained in the period of 10 to 12 weeks after transection when fish could endure the high speed for almost 15 min. The aerobic capacity of muscle fibers in distal parts of the body was not strongly affected by the lesion. The only important change in aerobic capacity was observed in the neuropil of distal parts of the cords where, at three months after transection, NADH-TR activity was increased to approximately 150% of control values. On the basis of our findings, we assume that it is not the condition of the neuromuscular system, but rather a deficient co-ordination between proximal and distal body parts of lesioned fish that accounts for the relatively poor performances in endurance tests. Furthermore, differences in timing of improvements in swimming at 2 and 4.5 BL/s indicate that the spinal circuitries serving the slow parts of the neuromuscular system recover at an earlier stage than those serving the fast parts.