Background <p>Bariatric surgery presents a significant alleviation for non-alcoholic fatty liver disease (NAFLD), which relies in part on achieving substantial weight loss in post-surgical period.&#xa0;We aimed to understand the effect of bariatric surgery on NAFLD remission via metabolomics and to validate the results in a general population-based cohort.</p> Methods <p>In a pilot study, ten patients with NAFLD who underwent bariatric surgery were enrolled. The remission of hepatic steatosis was assessed by MRI-derived proton density fat fraction (PDFF) before and 3-month after surgery. Temporal associations of body mass index (BMI) reduction, alteration in metabolomic biomarkers, and NAFLD remission were quantified by using cross-lagged models, which were then validated in a general population-based cohort (<i>n</i> = 1258).</p> Results <p>At 3-month after surgery, BMI reduction of 6.9 (SD 1.9) kg/m² and MRI-PDFF reduction of 9.6% (5.4) (all <i>p-</i>value&#xa0;&lt; 0.001) were achieved. Of the 64 metabolomic biomarkers quantified, 19 biomarkers showed significant differences between pre- and post-surgery (false discovery rate-corrected <i>p-</i>value&#xa0;&lt; 0.05). Temporal associations were observed between BMI reduction and 5 metabolomic biomarkers, while 3 (chenodeoxycholic acid [CDCA], palmitoylcarnitine, and hippuric acid) were further validated in the general population-based cohort. CDCA was able to explain 18% of the association between BMI reduction and NAFLD remission (<i>p-</i>value&#xa0;&lt; 0.05). In the general population-based cohort, Mendelian randomization showed that genetically elevated CDCA level was associated with a higher risk of liver fibrosis.</p> Conclusions <p>CDCA is a potential mediator and may predict long-term surgical benefits in liver fibrosis regression.</p>