Abstract: The model study of moisture measurement in meat by LF-NMR was studied in this paper. The fresh pork sample was selected by MesoMR23 low field nuclear magnetic analysis instrument to determine the T₂ relaxation characteristics. At the same time,the direct moisture method was used to determine the actual moisture content in the meat. Multiplication,partial least squares(PLSR)and principal component regression(PCR)were used to establish prediction models. Results indicated that the decision coefficients R² of the three prediction models were all greater than 0.9. LSE,PLSR and PCR,the coefficient of determination between the predicted value of the sample moisture content and the reference value were 0.960,0.969 and 0.941,respectively,and the predicted root mean square deviations were 0.048,0.048 and 0.104,respectively. Therefore,the PLSR model has better prediction results.