A first approach to apply methods of inertial navigation for a motion tracking in a real-scale snow avalanche is presented. In the course of an artificially triggered avalanche event near the Flüelapass in Switzerland in January 2013, in situ inertial measurements (accelerations and angular velocities) have been combined with supplementary velocity measurements by means of Global Positioning System (GPS) and Doppler radar. Specifically, a commercial 9-axis Inertial Measurement Unit (IMU) has been mounted inside a rigid sphere and placed in the avalanche track such that it was entrained by the moving snow. To determine velocity and position, standard methods of state estimation (like Kalman filtering) are not optimally suited for this specific case. Instead, a new time integration algorithm is applied. The procedure turns out to be stable over a time period of about 4 s, which corresponds to a path length of about 65 m. The obtained velocity evolution fits well with the corresponding GPS and radar results. A comparison of the derived IMU trajectory with the corresponding topography reveals the reliability of position data within this time period. The spatial resolution of the IMU trajectory is such that a vertical motion of the sensor unit relatively to the terrain can be detected.