Non-alcoholic fatty liver disease (NAFLD) is common in diabetes and obesity but a minority only have clinically significant liver fibrosis. Improved non-invasive risk assessment is needed as the most commonly used clinical risk algorithm, the NAFLD Fibrosis Score (NFS), classifies many patients as having ‘intermediate-risk’ of severe fibrosis, who do not have significant fibrosis on subsequent biopsy. As fibroblast activation protein (FAP) is most similar in structure to dipeptidyl peptidase-4 but with expression limited in adults to pathological sites, including in liver fibrosis, it is conceptually a good candidate biomarker. Aims: To determine whether circulating levels of FAP (cFAP) associate with clinically significant liver fibrosis in NAFLD and conversely, if low cFAP has value in excluding significant fibrosis, particularly combined with NFS. Methods: cFAP was measured in 106 with type 2 diabetes who had transient elastography (Cohort 1) and 146 with morbid obesity who had liver biopsy (Cohort 2) to exclude significant liver fibrosis. Results: In Cohort 1, cFAP was an independent risk marker for median liver stiffness (LSM) ≥10.3 kPa (consistent with severe fibrosis) with OR (per SD increase) of 2.0 (95% CI 1.2-3.4), p=0.006. Conversely, there was 0.12 OR (95% CI 0.03-0.61) of LSM ≥10.3kPa for those in the lowest compared with the highest FAP tertile (p=0.010). cFAP levels below 730pmolAMC/min/mL had 95% NPV for severe fibrosis. Most important, low cFAP reclassified 41% of 64 subjects from ‘intermediate-risk’ by NFS, to ‘low-risk’. In Cohort 2, per SD increase in cFAP, there was 1.7 fold (95% CI 1.1-2.8) increased odds of significant fibrosis (F≥2), p=0.021, and low cFAP correctly reclassified 49% of 73 subjects from ‘intermediate-risk’ of significant fibrosis by NFS, to ‘low-risk’. Conclusions: cFAP co-segregates with liver fibrosis in NAFLD and lower cFAP, used after NFS, has clinical utility in excluding significant fibrosis in type 2 diabetes and obesity.