The practicality of using restrained as well as standard isoinertial strength tests in predicting the maximum acceptable weight of lift (MAWL) of Chinese male subjects was examined using a restrained strength test, which allows the subject to pull on the load in front of the body and to apply force in a functional free posture. Prediction models were constructed and evaluated under task conditions of two lifting ranges, two box sizes, and three lifting frequencies. The following measurements were taken for each subject in each test: (1) the body posture and joint angles, and (2) electromyographic activities of the arm muscles (biceps brachii), back muscles (erector spinae), and leg muscles (rectus femoris). Results indicate that the restrained knuckle height isoinertial strength (RK) was related significantly to the MAWL for all tasks, and the correlation coefficients ranged from 0.42 to 0.69. A step-wise multiple regression analysis based on data from the 22 subjects showed that RK, chest circumference, and thigh length accounted for 90 to 94% of the variance. A validation effort on a different sample (N = 10) showed that the MAWLs can be predicted within an absolute mean (SD) error of 0.84 (3.30) kg. Percentages of muscle activation levels indicated that, other than strength from the erector spinae, the strongest contribution to the RK value was that of biceps brachii strength. Since the strength of the body's upper extremity was also recruited, the weak upper extremity strength of Chinese subjects was better reflected. Evidence for the existence of a close match between MAWL and RK values, and the conditions of the existence, suggest that RK is good predictor for the MAWL.