The aim of this study was to examine the predictive scoring system of advanced liver fibrosis in severely obese Japanese patients. Seventy-two patients underwent laparoscopic sleeve gastrectomy and intraoperative liver biopsies. We classified these patients into two groups: Brunt stage ≥ 2 (advanced fibrosis) and 0/1 (none/mild fibrosis). A logistic regression analysis was performed to identify the predictors of advanced fibrosis. Sixteen patients had advanced fibrosis, while 56 had no/mild fibrosis. The prevalence of type 2 diabetes mellitus (T2DM) in advanced fibrosis group was significantly higher than in none/mild fibrosis. An univariate analysis of the factors predicting advanced fibrosis showed significant differences in AST/ALT ratio, serum insulin levels, HOMA-IR, and type IV collagen 7S in the T2DM group. According to a multivariate analysis, type IV collagen 7S was an independent predictor and the cutoff value was 5.6ng/mL. We created a flow chart; high risk (T2DM and type IV collagen 7S ≥ 5.6ng/mL), moderate risk (T2DM and type IV collagen 7S < 5.6ng/mL), and low risk (non-DM). For those at high risk, the sensitivity, specificity, positive predictive value, and negative predictive value were 56.2%, 94.4%, 75.0%, and 87.9%, respectively. This classification system has the potential to accurately categorize the risk of liver fibrosis.