A new algorithm has been developed to non-invasively identify patients with nonalcoholic steatohepatitis (NASH) at risk of progression to cirrhosis. Using a derivation cohort of 350 patients with suspected nonalcoholic fatty liver disease, the best-fitting multivariable logistic regression model predicting NASH was established, using measurements of liver stiffness by FibroScan and serum aspartate aminotransferase (AST) levels. This score, named FAST (FibroScan–AST), performed well in both the derivation dataset as well as in external validation cohorts (n = 1,026), and could reduce unnecessary liver biopsies.