The quality of man-machine interface (MMI) directly affects efficiency, safety and comfort of the man-machine system. Therefore, it is significant to study an appropriate evaluation technology for MMI. The method of grey relational analysis based on interval numbers is put forward in the field of MMI evaluation in this paper, based on human thinking having the grey and fuzzy characteristic. In this method, interval numbers are employed as a mean to reflect the rating of indexes instead of real numbers, which offer uncertain information. By judging the correlative degree between the evaluation criterion and the design schemes, the optimal scheme can be found. Based on the proposed method, a novel subjective evaluation model of MMI is set up and applied in the subjective evaluation on the digital control system console of the power plant. Results of the evaluation example prove that the proposed method for MMI evaluation is reasonable and applicable.