In crop production, the unreasonable use of chemical pesticides to prevent from plant diseases is widespread, which affects product quality and food safety seriously. Therefore, it is of great significance to identify plant disease quickly and adopt appropriate control measures to improve the quality of crops. In this paper, the healthy leaves, rust spot and green area near the spot of rust diseased leaves of broad bean, corn, allium fistulosum and garlic were studied by a tri-step IR spectroscopy method, including Fourier transform infrared (FT-IR) spectroscopy, second derivatives infrared (SD-IR) spectroscopy and two-dimensional correlation infrared (2D-IR) spectroscopy. The results showed that tiny differences were observed in the intensities and shape of several peaks in the original spectra of each crop leaves. And several peak intensity ratios in the original spectra were different. The peak intensity ratio A1 410/A1 646 of the healthy leaves, green area near spot and rust spot of rust diseased leaves of broad bean were 0.698, 0.624 and 0.616 respectively, and the corresponding ratio A2 926/A1 646 were 0.665, 0.638 and 0.552 respectively. The corresponding ratio A1 649/A1 055 of corn were 0.813, 0.696 and 0.691 respectively, and the corresponding ratio A1 382/A1 055 were 0.552, 0.478 and 0.465 respectively; the corresponding ratio of A2 926/A1 055 were 0.574, 0.467 and 0.469 respectively. The corresponding ratio A1 382/A1 061 of allium fistulosum were 0.843, 0.821 and 0.704 respectively; the corresponding ratio of A2 923/A1 061 were 0.707, 0.680 and 0.489 respectively. It can be seen that the intensity ratios of the rust spot and the green area near the spot of rust leaves were lower than that of the healthy leaves. More significant differences were exhibited in their SD-IR spectra in the range of 1 800～800 cm-1, and clearer differences in the position and intensity of auto and cross peaks were observed in the range of 860～1 690 cm-1 in 2D synchronous correlation spectra. The healthy leaves of broad beans showed 4 strong auto-peaks and 2 strong positive cross peaks, and 5 strong auto-peaks and 4 strong positive cross peaks were revealed in the green area near the spot of rust diseased leaves, and 2 strongest auto-peaks, 5 medium strong auto peaks and 5 strong positive cross peaks were appeared in the rust spot of rust diseased leaves. The intensity of auto-peaks of the rust spot of broad bean rust leaves were the strongest while the intensity of auto-peaks of the healthy leaves were the weakest. There were 9 strong auto-peaks and 12 strong positive cross peaks in the healthy leaves of corn, and 11 strong auto-peaks, 3 strongest positive cross peaks and 11 medium strong positive cross peaks in the green area near the rust spot on the diseased leaves, and 6 strong auto-peaks and 3 strong positive cross peaks in the rust spot of rust diseased leaves. There were 9 strong auto-peaks and 8 strong positive cross peaks in the healthy leaves of garlic, and 2 strongest auto-peaks, 9 medium strong auto-peak and 10 strong positive cross peaks appeared in the green area near spot of rust diseased leaves, and 6 strong auto-peaks and 1 strong positive cross peaks in the rust spot of rust diseased leaves. The intensity of auto-peaks of the green area near the spot of rust diseased leaves of corn and garlic were the strongest while the intensity of auto-peaks of the rust spot of rust diseased leaves were the weakest. There were 9 strong auto-peaks and 5 strong positive cross peaks in the healthy leaves of allium fistulosum, and 8 strongest auto-peaks and 3 strong positive cross peaks in the green area near the spot of rust diseased leaves, and 3 strong auto-peaks in the rust spot of rust diseased leaves. The auto-peaks of the healthy leaves of allium fistulosum were the strongest, while the auto-peaks of the rust spot of the rust leaves were the weakest. It is demonstrated that FT-IR combined with SD-IR and 2D-IR spectroscopy could be used to discriminate the crops rust leaves rapidly and effectively. Tri-step IR spectroscopy might provide a spectral method for detecting crop disease.