In　our　country,　a　certain　type　of　accelerometer　is　stored　in　four　typical　climate　zones　which　are　sub　humid,　Asia　hot,　mild　and　dry　zones.　In　view　of　the　historical　statistics　(accelerometer＇s　success　or　failure,　and　incomplete　failure　data),　assuming　that　they　are　subject　to　exponential　distribution,　Weibull　distribution,　extreme　value　distribution　and　logarithmic　normal　distribution,　and　by　combining　with　the　engineering　data　to　handle　the　exceptions　and　＇upside　down＇　data,　a　minimum　χ2　estimation　method　is　used　to　estimate　the　parameters　of　the　distribution　function,　and　the　fit　goodness　of　＇obey　distributed　random　variables　under　different　degrees　of　freedom　χ2＇　test　is　calculated.　In　this　way　the　storage　reliability　distribution　function　of　an　accelerometer　is　obtained.　The　fault　data　statistics　verify　the　applicability　and　validity　of　the　proposed　mathematical　model　assessment　method,　and　the　accelerometer　natural　storage　life　is　obtained　under　the　condition　of　confidence　0.90　and　reliability　0.95.