[Objective] This study aims to isolate Clostridium strains from Luzhou-flavor baijiu pit mud and evaluate their metabolic characteristics. [Methods] 16S rRNA gene sequencing was used to analyze the clostridial community structure in pit mud; specific media were predicted according to the sequences of high-abundance OTUs scattered in Clostridia using KOMODO database; volatile compounds of pit mud and the representative isolated Clostridium strains' fermentation broth were detected by headspace solid phase microextraction combined with gas chromatography-mass spectrometry. [Results] Thirty-one strains belonging to 14 species of genus Clostridium were isolated using the 7 media predicted by KOMODO database. Based on the analysis of volatile metabolites, these Clostridium strains could be divided into two clusters, including cluster I, i.e. strains mainly producing acids (C. carboxidivorans, C. sporogenes and C. tyrobutyricum) and cluster II, i.e. strains mainly producing alcohols (C. beijerinckii, C. butyricum and C. sphenoides). [Conclusion] KOMODO method is helpful for the isolation of Clostridium or other bacteria from the pit mud of Luzhou-flavor baijiu, and elucidating their metabolic profiles could be conductive to analyze the function of brewing microbes in baijiu flavor formation.