【Objective】 To diagnose citrus Huanglongbing(HLB) timely to prevent citrus production from the spread of the disease. 【Method】 A detection method of citrus HLB based on modulation chlorophyll fluorescence measurements was investigated. Fluorescence parameters were extracted from MINI-PAM, and analyzed by probability neural network (PNN) model and classification to distinguish among healthy citrus, HLB-infected citrus and etiolated citrus due to non-HLB problems. 【Result】 The average detection accuracy for different classes of citrus symptoms was above 76.93%, and that for some classes even reached 100%. 【Conclusion】 It is feasible to use the modulation chlorophyll fluorescence measurement combined with PNN model to detect citrus HLB.