Abstract: To solve the problem in first-round loading of a dual-purpose machine gun, the loading test was carried out by using a linear displacement sensor, and the importance factors which affect the first-round loading were revealed, i.e. two maximum values of resistances transferred from feed slide to the loading handle. Five points were taken in two segments of big-lever curve respectively, and the relationship between two maximum values and interpolation points were found out by using cubic spline curve fitting. Taken these two values as objective functions and the Y coordinates of interpolation points as the optimized parameters, the big-lever curves were optimized by using genetic algorithm, and the objective functions were reduced by 47.24% and 39.3%. The method can provide reference for the design and improvement of the similar weapons.