Abstract: A two-factor three-level full-factor hot air drying experiment was carried out at different drying temperatures（50, 60, 70 °C）and pellet thicknesses（4.24, 9.15, 15.52 mm）, in order to explore drying characteristics and moisture migration law of gluten pellets． Low-field nuclear magnetic resonance technology was used to analyze moisture migration law in drying process, and a drying kinetic model and moisture prediction model was established． Rresults showed that drying rate and moisture ratio of gluten pellets decreased significantly with increase of temperature（P<0.05）． Effective water diffusion coefficient increaseed with temperature and pellet thickness raising． Calculation results of determination coefficient（R²）, sum of squares of deviation（\begindocument \chi ^2 \enddocument）and root mean square error（RMSE）show that Modified Page thin layer drying model had high fitting accuracy to drying test data of gluten pellets． And regression model of model parameters（k,n）, drying temperature（T）and pellet thickness（H）was established（R²＞0.926）． Low-field nuclear magnetic resonance transverse relaxation time（T₂）inversion spectrum showed that peak area of each moisture gradually decreased, and peak position gradually approached bound water with increase of drying time． Relationship between moisture content（M）and drying time（t）, pellet thickness（H）, drying temperature（T）, total peak area（A）of relaxation inversion spectrum was established． Results showed that prediction accuracy was high（R²=0.933）． Research results could provide reference for drying process of gluten pellets．