When the modal model is going to be used for structural modification or for structural response simulation, the scaled mode shapes must be known. If natural input modal analysis is performed, only un-scaled mode shapes can be obtained and an extra method is necessary to obtain the scaling factor. In this paper, two new methods based on mass change are proposed. The first method involves small mass changes in two repeated tests allowing to achieve good accuracy. The second method involves only one mass change and enables the scaling factors of both the modified and unmodified mode shapes to be obtained. Finally, the effect of the normalization used in the mode shapes and the accuracy of each method are analyzed by simulation. ; When the modal model is going to be used for structural modification or for structural response simulation, the scaled mode shapes must be known. If natural input modal analysis is performed, only un-scaled mode shapes can be obtained and an extra method is necessary to obtain the scaling factor. In this paper, two new methods based on mass change are proposed. The first method involves small mass changes in two repeated tests allowing to achieve good accuracy. The second method involves only one mass change and enables the scaling factors of both the modified and unmodified mode shapes to be obtained. Finally, the effect of the normalization used in the mode shapes and the accuracy of each method are analyzed by simulation.