Fish novirhabdoviruses, including viral hemorrhagic septicemia virus (VHSV), hirame rhabdovirus (HIRRV), and infectious hematopoietic necrosis virus (IHNV), harbor a unique non-virion (NV) gene that is crucial for efficient replication and pathogenicity. The effective levels and the function of the N-terminal region of the NV protein, however, remain poorly understood. In the present study, several recombinant VHSVs, which completely lack (rVHSV-ΔNV) or harbor an additional (rVHSV-dNV) NV gene, were generated using reverse genetics. To confirm the function of the N-terminal region of the NV protein, recombinant VHSVs with the NV gene that gradually mutated from the start codon (ATG) to the stop codon (TGA), expressed as N-terminally truncated NV proteins (rVHSV-NV1, -NV2, and -NV3), were generated. CPE progression and viral growth analyses showed that epithelioma papulosum cyprini (EPC) cells infected with rVHSV-ΔNV or rVHSV-NV3-which did not express NV protein-rarely showed CPE and viral replication as opposed to EPC cells infected with rVHSV-wild. Interestingly, regardless of the presence of two NV genes in the rVHSV-dNV genome, EPC cells infected with rVHSV-dNV or rVHSV-A-EGFP (control) failed to induce CPE and viral replication. In EPC cells infected with rVHSV-dNV or rVHSV-A-EGFP, which harbored a longer VHSV genome than the wild-type, Mx gene expression levels, which were detected by luciferase activity assay, were particularly high; Mx gene expression levels were higher in EPC cells infected with rVHSV-ΔNV, -NV2, or -NV3 than in those infected with rVHSV-wild or rVHSV-NV1. The total amount of NV transcript produced in EPC cells infected with rVHSV-wild was much higher than that in EPC cells infected with rVHSV-dNV. However, the expression levels of the NV gene per viral particle were significantly higher in EPC cells infected with rVHSV-dNV than in cells infected with rVHSV-wild. These results suggest that the NV protein is an essential component in the inhibition of host type-I interferon (IFN) and the induction of viral replication. Most importantly, viral genome length might affect viral replication efficiency to a greater extent than does NV gene expression. In in vivo pathogenicity experiments, the cumulative mortality rates of olive flounder fingerlings infected with rVHSV-dNV or rVHSV-wild were similar (60-70%), while those of fingerlings infected with rVHSV-A-EGFP were lower. Moreover, the virulence of rVHSV-ΔNV and rVHSV, both harboring a truncated NV gene (rVHSV-NV1, -NV2, and -NV3), was completely attenuated in the olive flounder. These results suggest that viral pathogenicity is affected by the viral replication rate and NV gene expression. In conclusion, the genome length and NV gene (particularly the N-terminal region) expression of VHSVs are closely associated with viral replication in host type-I IFN response and the viral pathogenicity.