Pathogenic microorganisms are one of the main environmental factors causing biological diseases, and their rapid spread and variety may lead to the outbreak of animal infectious diseases or plant diseases. However, the air environment is complex, the pathogen fungal infectious disease outbreak early concentration is low and the shape and size is small. There is a lack of rapid, accurate, portable and low-cost early detection methods for pathogenic fungi. Here, we construct a multispectral diffraction microfluidic sensing system for pathogenic fungi. Firstly, a microfluidic chip with the function of targeted purification of pathogenic fungi was designed, and the diffraction fingerprint of purified pathogenic fungi was collected on the chip. By analyzing the light diffraction spectrum, the morphological and absorption characteristics of different fungi at different wavelengths were obtained, and different kinds of pathogenic microorganisms in complex environment were identified. Then, seven weighting schemes were compared to select the characteristic wavelength that best reflects spore differences. Dispersion intensity compared of fingerprint (DICF) and Relative superimposed light intensity (RSLI) were defined as training indexes. To obtain higher recognition accuracy, two parameters of cluster centroid distance and dispersion are defined as evaluation indexes. Finally, 5 kinds of pathogenic fungi and particles with similar characteristics were selected for experimental verification, and the identification accuracy was 98.7%. We can foresee potential applications for this approach in disease control and environmental monitoring of airborne pathogens.