BACKGROUND & AIMS: Non-alcoholic steatohepatitis (NASH) is a leading cause of chronic liver disease in Western countries. Diagnosis of NASH requires a liver biopsy. We estimated the prevalence of NASH non-invasively in a population-based study using scores validated against liver histology. METHODS: Clinical characteristics, PNPLA3 genotype at rs738409, and serum cytokeratin 18 fragments were measured in 296 consecutive bariatric surgery patients who underwent a liver biopsy to discover and validate a NASH score (‘NASH score’). We also defined the cut-off for NASH for a previously validated NAFLD liver fat score to diagnose NASH in the same cohort (‘NASH liver fat score’). Both scores were validated in an Italian cohort comprising of 380, mainly non-bariatric surgery patients, who had undergone a liver biopsy for NASH. The cut-offs were utilized in the Finnish population-based D2D-study involving 2849 subjects (age 45-74 years) to estimate the population prevalence of NASH. RESULTS: The final ‘NASH Score’ model included PNPLA3 genotype, AST and fasting insulin. It predicted NASH with an AUROC 0.774 (0.709, 0.839) in Finns and 0.759 (0.711, 0.807) in Italians (NS). The AUROCs for ‘NASH liver fat score’ were 0.734 (0.664, 0.805) and 0.737 (0.687, 0.787), respectively. Using ‘NASH liver fat score’ and ‘NASH Score’, the prevalences of NASH in the D2D study were 4.2% (95% CI: 3.4, 5.0) and 6.0% (5.0, 6.9%). Sensitivity analysis was performed by taking into account stochastic false-positivity and false-negativity rates in a Bayesian model. This analysis yielded population prevalences of NASH of 3.1% (95% stimulation limits 0.2-6.8%) using ‘NASH liver fat score’ and 3.6% (0.2-7.7%) using ‘NASH Score’. CONCLUSIONS: The population prevalence of NASH in 45-74 year old Finnish subjects is approximately 5%.