A new measurement system for the complete motion analysis of dummy heads during a crash test was developed. The approach uses inertial measurement supported by standard photogrammetric methods. Once initialized, an INS (inertial navigation system) requires no external references in order to determine its position, orientation, or velocity. Careful selection of sensors and their calibration play a fundamental role in applying inertial measurement for vehicle impact tests. A suited sensor unit was presented together with a data acquisition system. A static multi-position calibration method in combination with a centrifuge test bench was chosen. Fully automated calibration guaranties high performance and stability of derived parameters. A pendulum test bench with several degrees of freedom was constructed which generates dynamic test sequences. It allows the entire evaluation of the INS under crash-like conditions. Performance analyses using the pendulum test bench have shown that motion tracking with position errors smaller than 3 mm is achievable. Furthermore, an algorithm was developed that initializes the INS from photogrammetric data. Only few supporting points of the trajectory must be given at the beginning and at the end of the measurement. A standard set of two cameras is sufficient in generating all required reference data. Measurement uncertainty of initial values can be rejected. Several pendulum tests have been conducted for an analytical investigation of the INS. Marginal position errors between 2 and 4 mm have been achieved. Furthermore, a sled test was done. In this case, photogrammetry was used as a reference thus making a performance analysis of the INS possible. The maximum position error was amounted to 3 mm. Finally, the results of a real crash test clearly demonstrate the performance of the new measurement procedure. A position error of less than 4 mm over a period of 200 ms was achieved using standard photogrammetric equipment.