Abstract. Citrus huanglongbing disease is an important disease of citrus species, which leads to the change of chemical composition in leaves. The infrared technique may be advantageous to characterize the chemical differences between Huanglongbing infected and healthy leaves rapidly and nondestructively. In this study, the near-infrared spectra of citrus leaves were obtained in the field by using MicroNIR 1700 Spectrometer, and then precision detection of citrus HLB-associated bacteria (Candidatus Liberibacter asiaticus, CLas) by RT-qPCR for verification. The method of Soft Independent Modeling of Class Analogy(SIMCA) and Partial Least Squares Discriminant Analysis (PLS-DA) was used to establish the qualitative discriminant models of HLB diagnosis, and using first derivative and Savitzky-Golay for further data processing. The results showed that the correct rejection rate of Principal Component Analyses (PCA) model of CLas-negative leaves was more than 80%, and correct recognition rate of CLas-positive leaves was more than 95% in different citrus orchards from the field. The correct rejection rate of PLS-DA model of CLas-negative leaves was more than 83%, and correct recognition rate of CLas-positive leaves was more than 99% in different citrus orchard from the field. Both models were well distinguished whether the citrus leaves were CLas-negative or CLas-positive(even if the citrus leaves was asymptomatic), which provided a new method for the rapid diagnosis and early warning of HLB in the field.