Transient elastography (TE) is accurate in staging fibrosis noninvasively. However, a reliable serum biomarker with comparable accuracy is also important, especially when TE is unreliable/unavailable. Therefore, we aimed to evaluate the diagnostic performance of serum Golgi protein 73 (GP73) for significant fibrosis in patients with chronic HBV infection. A total of 801 patients with chronic liver disease (CLD; 492 chronic HBV infection and 309 non-HBV liver disease) with liver biopsy performance were enrolled. Healthy controls (n = 180) and hepatocellular carcinoma (HCC) patients (n = 85) were included for comparisons. Liver biopsy was used as the reference method for fibrosis staging. Serum GP73 level was measured in duplicate in double-blind fashion. Serum GP73 was highest in HCC but also significantly higher in chronic hepatitis B than in healthy controls. The elevation of serum GP73 in non-HCC patients was significantly associated with the presence of significant fibrosis independently of ALT level, liver stiffness (LS) value, inflammation grade and other confounding factors. The diagnostic performance of serum GP73 was accurate in antiviral-naïve HBV patients (area under the receiver operating curve [AUROC], 0.76 95% CI: 0.72-0.81) but not in patients with ongoing antiviral treatment (AUROC, 0.60). The utility of serum GP73 was also confirmed in non-HBV CLD (AUROC, 0.80 95% CI: 0.75-0.85). Serum GP73 was comparable to LS (AUROC, 0.78 95% CI: 0.73-0.82) and significantly better than AST to platelet ratio index (APRI) (AUROC, 0.67 95% CI: 0.62-0.72) and FIB-4 (AUROC, 0.68 95% CI: 0.63-0.73). In conclusion, serum GP73 is an accurate serum marker for significant fibrosis in chronic HBV infection, with higher accuracy than APRI and FIB-4. Serum GP73 is potentially a complementary tool for TE when evaluating the necessity of antiviral treatment, particularly in patients without definite antiviral indication.