Bioartificial livers may act as a promising therapy for fulminant hepatic failure (FHF) with better accessibility and less injury compared to orthotopic liver transplantation. This study aims to evaluate the efficacy and safety of a fluidized bed bioartificial liver (FBBAL) and to explore its therapeutic mechanisms based on metabolomics. FHF was induced by D-galactosamine. Eighteen hours later, pigs were treated with an FBBAL containing encapsulated primary porcine hepatocytes (B group), with a sham FBBAL (containing cell-free capsules, S group) or with only intensive care (C group) for 6 h. Serum samples were assayed using ultra-performance liquid chromatography-mass spectrometry. The difference in survival time (51.6 ± 7.9 h vs. 49.3 ± 6.6 h) and serum metabolome was negligible between the S and C groups, whereas FBBAL treatment significantly prolonged survival time (70.4 ± 11.5h, P < 0.01) and perturbed the serum metabolome, resulting in a marked decrease in phosphatidylcholines, lysophosphatidylcholines, sphingomyelinase, and fatty acids and an increase in conjugated bile acids. The FBBAL exhibits some liver functions and may exert its therapeutic effect by altering the serum metabolome of FHF pigs. Moreover, alginate-chitosan capsules have less influence on serum metabolites. Nevertheless, the alterations were not universally beneficial, revealing that much should be done to improve the FBBAL.