Perineuronal nets (PNNs) surrounding neuronal cell bodies regulate neuronal plasticity during development, but their roles in regeneration are unclear. In the PNNs, chondroitin sulfate (CS) is assumed to be involved in inhibiting contact formation. Here, we examined CS expression in PNNs in the ventral horn of a goldfish hemisected spinal cord in which descending axons regenerate beyond the lesion to connect with distal spinal neurons. In intact fish, chondroitin sulfate A (CS-A)–positive PNNs accounted for 5.0% of HuC/D-immunoreactive neurons, and 48% of choline acetyltransferase (ChAT)-immunoreactive neurons. At 2, 4 and 8 weeks after spinal hemisection, CS-A–positive PNNs accounted for 8.4%–9.9% of HuC/D-immunoreactive neurons, and 50%–60% of ChAT-immunoreactive neurons, which was not significantly different from intact fish. Chondroitin sulfate C (CS-C)–positive PNNs accounted for 6.4% of HuC/D-immunoreactive neuron, and 67% of ChAT-immunoreactive neurons in intact fish. At 2, 4 and 8 weeks after spinal hemisection, CS-C–positive PNNs accounted for 7.9%, 5.5% and 4.3%, respectively, of HuC/D-immunoreactive neurons, and 65%, 52% and 42%, respectively, of ChAT-immunoreactive neurons, demonstrating a significant decrease at 4 and 8 weeks after spinal hemisection. Among ventral horn neurons that received descending axons labeled with tetramethylrhodamine dextran amine (RDA) applied at the level of the first spinal nerve, CS-A–positive PNNs accounted for 53% of HuC/D-immunoreactive neurons. At 2 and 4 weeks after spinal hemisection, CS-A–positive PNNs accounted for 57% and 56% of HuC/D-immunoreactive neurons, which was not significantly different from intact fish. CS-C–positive PNNs, accounted for 48% of HuC/D-immunoreactive neurons that received RDA-labeled axons. At 2 and 4 weeks after spinal hemisection, CS-C–positive PNNs significantly decreased to 22% of the HuC/D-immunoreactive neurons, and by 4 weeks after spinal hemisection they had returned to 47%. These findings suggest that CS expression is maintained in the PNNs after spinal cord lesion, and that the descending axons regenerate to preferentially terminate on neurons not covered with CS-C–positive PNNs. Therefore, CS-C in the PNNs possibly inhibits new contact with descending axons, and plasticity in the spinal neurons might be endowed by downregulation of CS-C in the PNNs in the regeneration process after spinal hemisection in goldfish.