For　monitoring　and　modeling　a　structural　element＇s　or　system＇s　lifecycle　behavior,　this　paper　presents　a　new　model　of　non-stationary　Markov　Chain　as　a　random　process.　It　is　to　take　into　account　the　element＇s　or　system＇s　age　as　an　important　factor　in　modeling.　The　transition　probability　matrix　is　now　age-dependent　not　constant　for　all　ages　as　in　a　stationary　Markov　Chain.　Therefore,　a　number　of　such　matrices　will　be　needed　to　model　the　entire　life　of　the　object＇s　deterioration　and　renewal　process.　Given　inspection　data　of　condition　state　evolution,　a　fitting　scheme　is　also　proposed　here　to　estimate　the　non-stationary　transition　probabilities　depending　on　age.　Application　of　the　proposed　model　to　Michigan　bridge　inspection　data　shows　improved　modeling　results.　This　approach　was　also　applied　to　the　bridge　condition　rating　data　in　the　US　National　Bridge　Inventory　format　and　thereby　produced　more　realistic　predictions.　©　2011　Taylor　&　Francis　Group,　London.