For molding equipment, the accurate measurement of tie-bar stress helps improve product quality and prolong the lifespan of the machine. In our previous work, the mathematical model between the ultrasonic time difference (UTD) and the tie-bar stress was established, and this model was shown to have high accuracy with the maximum of difference square being only 1.5678 (MPa)2. Nevertheless, when calculating the ultrasonic time difference using the cross-correlation function (CCF) method, it is necessary to manually divide the position of the echoes. This makes it impossible to automate the stress measurement and it is incompetent when dealing with massive data. In this paper, a non-intrusive (N-I) ultrasonic measurement method for tie-bar stress based on the minimum entropy blind deconvolution is firstly proposed. The results of simulation and verification experiments show that the proposed method has high precision, and the maximum relative error is only 2.4392%. Hence this method has been successfully applied in the tie-bar stress measurement in molding equipment. Compared with the CCF method, the accuracy is improved and the maximum relative error is reduced from 3.67% to 1.68% when the clamping force is small. In summary, the proposed method is comparable to the CCF method in accuracy, and has a higher resolution during echos disturbance. Moreover, this approach does not require human intervention and is helpful to realize the automation of the tie-bar stress measurement for the molding equipment.