A method for controlling the operation of a reversible belt retractor to release a belt extraction lock, which can be activated by an acceleration sensor, so that, after the belt retractor has been triggered as a consequence of a hazardous situation having been detected and after the hazardous situation has been recognized as being over, the operation of the belt retractor is controlled by a release signal at a release time in order to bring about the release of the belt extraction lock to shift it from a blocking state into a comfort-providing state, comprising:
- determining a release time in which the belt extraction lock is not effective with a sensor model algorithm based on a model of the acceleration sensor and with at least one variable characterizing running dynamics, and
- controlling the operation of the belt retractor by the release signal at the release time.