Nuclear magnetic resonance （NMR） technology combined with the chemometrics method was developed for the analysis on the difference of NMR hydrogen spectrum of four kinds of floral origins containing acacia honey， vitex honey， sunflower honey and Manuka honey. Pre-processing method was optimized and the NMR data were preprocessed effectively， then the multivariate statistical analysis methods containing principal component analysis（PCA）， partial least squares discriminant analysis （PLS-DA） and orthogonal partial least squares discriminant analysis （OPLS-DA） were used to extract the classification information of each group based on the nuclear magnetic signals. The results showed that disperse liquid-liquid micro-extraction （DLLME） could avoid the signals of low contents of characteristic compounds covered by the large amounts of sugar in honey and the signals of chemical shifts from δ=0.5-10.0 was observed. The data was preprocessed by log transformation and Pareto scaling， and four kinds of unifloral-honeys could be distinguished by the PCA model. The explanation and prediction abilities based on the PLS-DA model were 93.2% and 87.6%， respectively. The permutation test was used to validate the model externally and showed that model was not fitted and was robust. Corresponding NMR chemical shifts several significant markers were identified as the characteristic variables to distinguish honeys from the loading plots and correlation coefficient analysis of OPLS-DA model. The proposed method was simple and fast， and could be extended to distinguish and identify honeys from different floral origins， and provide effective method and prediction model for quality evaluation of different unifloral-honeys.