To improve protein delivery to the CNS following intracerebroventricular administration, we compared the distribution of a human Cu/Zn superoxide dismutase:tetanus toxin fragment C fusion protein (SOD1:TTC) in mouse brain and spinal cord with that of tetanus toxin fragment C (TTC) or human SOD1 (hSOD1) alone, following continuous infusion into the lateral ventricle. Mice infused with TTC or SOD1:TTC showed intense anti-TTC or anti-hSOD1 labeling, respectively, throughout the CNS. In contrast, animals treated with hSOD1 revealed moderate staining in periventricular tissues. In spinal cord sections from animals infused with SOD1:TTC, the fusion protein was found in neuron nuclear antigen-positive (NeuN+) neurons and not glial fibrillary acidic protein-positive (GFAP+) astrocytes. The percentage of NeuN+ ventral horn cells that were co-labeled with hSOD1 antibody was greater in mice treated with SOD1:TTC (cervical cord = 73 ± 8.5%; lumbar cord = 62 ± 7.7%) than in mice treated with hSOD1 alone (cervical cord = 15 ± 3.9%; lumbar cord = 27 ±4.7%). Enzyme-linked immunosorbent assay for hSOD1 further demonstrated that SOD1:TTC-infused mice had higher levels of immunoreactive hSOD1 in CNS tissue extracts than hSOD1-infused mice. Following 24 h of drug washout, tissue extracts from SOD1:TTC-treated mice still contained substantial amounts of hSOD1, while extracts from hSOD1-treated mice lacked detectable hSOD1. Immunoprecipitation of SOD1:TTC from these extracts using anti-TTC antibody revealed that the recovered fusion protein was structurally intact and enzymatically active. These results indicate that TTC may serve as a useful prototype for development as a non-viral vehicle for improving delivery of therapeutic proteins to the CNS.