To study a rapid and effective pork quality inspection method which could solve the problems of timeconsuming and high cost in traditional inspection methods, 109 groups of pork samples were collected from markets and farms. The samples were divided into healthy pork and diseased or dead pork by physical and chemical methods. A FT-NIR spectrometer was used to acquire the spectra of the samples. The spectra acquired were processed in noise reduction and dimensionality reduction. The results showed that after processing, a prediction model based on support vector machine algorithm was established. The accuracy of 5-fold cross validation was up to 94.5%. The results showed that the method could be used to classify the pork of different quality and provide the basis for the application of nearinfrared spectroscopy in pork quality detection.