Hepatitis G virus (HGV) is prevalent in patients with chronic liver disease and has been previously detected in liver specimens. However, it is unknown whether the virus is replicating in the liver or is simply a contaminant from serum. We sought to determine whether HGV was hepatotropic and to determine whether coinfection with HGV and hepatitis C virus (HCV) influenced the level of either virus. Virus was quantitated using branched DNA (bDNA) assay for both HGV and HCV in the liver explants and pretransplant serum samples from 30 transplant recipients: Group I, HGV/HCV coinfection (n = 10); group II, HCV infection alone, (n = 8); group III, HGV alone (n = 12). In patients with coinfection HCV (RNA) titers in liver were consistently higher than those for HGV RNA (median 1.13 x 10(8) and 360,000 Eq/g respectively, P < .01). The ratio of liver/serum viral RNA was significantly higher for HCV than for HGV (median 129 and 0.3 respectively, P < .01). Levels of HCV RNA were similar in patients with HCV infection alone versus those with HGV/HCV coinfection (median; liver = 1.15 x 10(7) vs. 1.13 x 10(8) Eq/g, serum = 500,000 vs. 200,000 Eq/mL) and levels of HGV RNA in liver and serum were similar in patients with HGV infection alone compared to those with HGV/HCV coinfection (median; liver = 1.2 x 10(6) vs. 4.0 x 10(5) Eq/g, serum = 4.5 x 106 vs. 2.6 x 10(6) Eq/mL). Levels of either virus appeared unaffected by the presence of an additional virus. The high ratio of HCV RNA levels in liver compared to serum is consistent with its known hepatotropism, but this pattern was not observed for HGV. The median liver/serum ratio of HGV RNA was less than unity, a finding consistent with serum contamination of liver tissue. Thus we conclude that the liver is not the main site of HGV replication.