The objectives of this study were twofold: to determine if the variability of dummy performance could be reduced by refinishing the joints and bearing surfaces of the dummy; and to establish a technique for preparing dummies for dynamic tests and measurements to be made for evaluating dummy performance. The testing was divided into three groups. The first phase evaluated the effect on data reproducibility by adjusting the dummy only before a complete series; the next phase evaluated the effect of adjusting before each run; the third phase evaluated the effect of changing the sled deceleration level. In each phase the lap belt, seat pan down, and seat pan forward loads were obtained. The data consist of graphs delineating the alteration on these loads during consecutive runs in each of the above phases. In addition, the loads were plotted versus the sled deceleration level and regression lines calculated. The results indicate that the standard errors about the regression lines of force versus deceleration were significantly decreased compared to previous studies. The seat pan forward force remained relatively constant if the dummy's joints were adjusted between each test. The dummy should have the joints adjusted every time or at least a minimum of every five tests.