Tapered roller bearings used to support pinion and differential gears in automotive drive axles perform best with accurate assembled preload. One of the most common high volume production assembly methods relies on bearing friction to adjust preload; however torque is an indirect measure of load, can be influenced by the raceway condition, and is difficult to control. A new method to measure preload utilizes frequency response to estimate axial preload from system stiffness. The stiffness can be measured dynamically and an assembly method automated to set preload without the need for torque or dimensional measurements. By eliminating the need for a torque signal, the raceway conditions which contribute to setting torque do not contribute to the preload setting accuracy. This study demonstrates the effectiveness of using frequency response to measure stiffness and estimate preload.