The phenolic and proline content were determined in honey samples of different floral origins (rapeseed, sunflower, buckwheat and Codonopsis) from five different regions of China. The phenolic and proline profile of these samples were used to construct a statistical model to distinguish honeys from different floral origins. Significant differences were identified among the studied honey samples from multivariate chemometric methods. The proline content varied among the four types of honeys, with the values decreasing in the order: buckwheat > Codonopsis > sunflower > rapeseed. Rapeseed honeys contained a high level of benzoic acid, while rutin, p-coumaric acid, p-hydroxybenzoic acid were present at relatively high levels in buckwheat honeys. Principal component analysis (PCA) revealed that rapeseed honey could be distinguished from the other three unifloral honeys, and benzoic acid, proline and kaempferol could serve as potential floral markers. Using 18 phenolic compounds and proline the honey samples were satisfactorily classified according to floral origin at 94% correct prediction by linear discriminant analysis (LDA). The results indicated that phenolic compounds and proline were useful for the identification of the floral origin of the four type honeys.