Beef is highly susceptible to microbial infection causing spoilage in the process of transportation, so the monitoring on beef spoilage is very important. This paper proved that beef in the process of spoilage released ammonia and carbon dioxide which were the main volatile substances. We used the long optical path FTIR spectra to detect the volatiles of beef spoilage. We quantitatively analyzed the change rule of ammonia and carbon dioxide in the process of beef spoilage to judge the state of beef. We used principal component analysis(PCA) to realize infrared spectral classification of volatile substance and accurately distinguish fresh and decayed beef. We used chemometrics methods: soft independent modeling cluster analysis(SIMCA) and partial least squares discriminant analysis(PLS-DA) to classify the characteristic spectrum of volatiles. The two methods both worked well. Results showed that the long optical path FTIR combined with chemometrics methods could distinguish fresh and decayed beef.