The genetic diversity of hepatitis G virus (HGV) was investigated. By using a RT-PCR procedure, 14% of either HBV (hepatitis B virus)- or HCV (hepatitis C virus)-positive Korean hepatitis patients were proved to be HGV positives. Nucleotide sequences in the E1 region of the eight isolates from Korean patients and the six previously reported isolates were compared. Nucleotide substitutions spread uniformly throughout the E1 region. Sequence homology among the Korean isolates was 84-99% and 88-99% at the nucleotide and amino acid sequences, respectively, whereas those from different geographic areas was slightly lower at both levels. At least two genotypes might exist among the Korean HGV isolates. Compared to the corresponding region of HCV, the E1 sequence from HGV is moderately conserved. In addition, as frameshift mutations were observed in most of the Korean isolates compared to the prototype HGV sequence, the Korean isolates might not use the translational initiation site of the prototype HGV for polyprotein translation. Because a putative signal sequence of E1 for entry into endoplasmic reticulum starts from the N-terminus of the polyprotein, and capsid-like peptides composed of basic amino acids could not be detected from the upstream region of E1, the core protein of HGV is absent, or at least not present, at the region next to 5'-UTR. Therefore, HGV could be clearly distinguished from other genera of Flaviviridae.