<p>The seat buffer mechanism (SBM) is a critical safety device for protecting astronauts during landing, requiring high reliability. Assessing SBM reliability is difficult due to limited sample sizes and complex data types. This paper proposes a direct reliability assessment method that evaluates the reliability confidence limit of SBM based on that of its subsystems. This method does not involve any approximate data conversion and thus improves assessment accuracy. Given that the SBM consists of metrological subsystems in series, a multiplication model for subsystem reliability confidence limits is established. The exact confidence limits of subsystem reliability are derived from metrological data. Based on the fiducial inference, a detailed derivation of the direct assessment method for SBM reliability is presented. Numerical studies show that this method surpasses traditional methods in coverage probability accuracy. An example illustrates that the proposed reliability model and assessment method for the SBM are reasonable and efficient.</p>