Aiming to explore the feasibility of the online nearinfrared quantitative analysis model of soybean meal from laboratory to factory transfer and application. Totally 117 soybean meal samples were collected nationwide and used, the nearinfrared quantitative analysis models of moisture and crude protein were established in the laboratory by partial least squares method, two kinds of nearinfrared online equipment installation methods were used in feed production enterprises. The laboratory models of nearinfrared quantitative analysis of soybean meal quality were transferred to feed production enterprises for online application by two different methods. The results showed that laboratory model can rapidly predict the content of moisture and crude protein in soybean meal (RSD was below 10%), R2P of moisture and crude protein were 0.83 and 0.86, RPD value of moisture and crude protein were 2.40 and 2.55, which meant that the model effect was fine; laboratory models were transferred to feed production enterprises for online analysis by different methods such as model correction and sample expansion, the deviation between the predicted value and the measured value was small and stable. The requirement for online detection can be achieved.