Frozen pork is an important national strategic materials in several countries, and it is crucial to monitor the freshness of frozen pork in real time. This study proposed a rapid and non-destructive detection method for the freshness of frozen pork without thawing based on portable fluorescence spectra system. Partial least squares regression (PLSR) models with optimal pre-treatment were developed based on frozen pork spectra of same temperature (−24 ℃) and different temperature (−14, −24, and −80 ℃) for total volatile basic nitrogen (TVB-N) and pH. The results showed that the model for frozen pork exhibited good prediction performance for these two indicators. For frozen pork at the same temperature (−24 ℃), the Rp and residual predictive deviation (RPD) of TVB-N were 0.9487 and 2.88, respectively. The Rp and RPD of pH were 0.9693 and 3.74, respectively. For frozen pork at different temperature (−14, −24, and −80 ℃), the Rp and RPD of TVB-N were 0.9151 and 2.28, respectively. The Rp and RPD of pH were 0.9703 and 3.80, respectively. The results demonstrated that portable fluorescence spectra can be used to assess the freshness of frozen pork without thawing.