There are many preferences expressing the quality of pork: color, pH, especially TVB-N content. Different quality pork has different spectral feature (in range of 400 to 1000nm). To detect quality attributes of pork easily, real-time, nondestructively, a portable device based on Vis/NIR spectral technique was developed. The device is mainly made up of four units: light source, spectrometer, controller and display screen. After hardware platform established, reflectance spectra of 44 samples were collected from this system. And their physicochemical characteristics such as color parameters, pH value and the content of total volatile basic-nitrogen (TVB-N) were measured in standard methods. Spectrum data acquired were processed by Savitzky-Golay filter(S-G) for noise removal, and then operated by standard normal variable transformation (SNV) for baseline drifts relieving. The partial least squares regression (PLSR) was used to build prediction models for L*, a*, b* pH* and TVB-N content, which could gain good prediction results with Rp of 0.92, 0.91, 0.92, 0.95 and 0.96 respectively. The results demonstrated that this device could be a promising tool applied to detecting pork quality attributes portably, real-time and nondestructively.