An LC-ESI-Q-TOF MS/MS method coupled with multivariate statistical analysis was developed for chemical fingerprinting of Cyathula officinalis Kuan. Ten peaks were selected as common peaks, and cyasterone was used as a reference. The relative areas of the common peaks were used for hierarchical clustering analysis, principal component analysis and similarity calculation. A total of 31 samples collected from different sources were classified into three groups (principal component analysis) or four groups (hierarchical clustering analysis). The similarities of the 31 batches of C. officinalis samples were between 0.653 and 0.999. The results obtained suggested that the chromatographic fingerprint technique could efficiently evaluate C. officinalis.