The　condition　assessment　of　oil–paper　insulation　is　a　vital　issue　in　transformer　management.　To　investigate　the　effect　of　moisture　content　on　the　accuracy　of　the　prediction　model,　accelerated　ageing　experiments　were　performed　with　different　initial　moisture　content　(IMC)　in　laboratory　condition.　Two　types　of　multi-parameter　lifetime　prediction　models,　e.g.　with　and　without　consideration　of　IMC,　were　presented　for　comparison　purpose.　The　authors　found　that　a　better　accuracy　for　its　prediction　results　could　be　achieved　by　establishing　the　prediction　models　with　the　consideration　of　IMC.　Besides,　models　considering　IMC　have　better　economic　feasibility　than　that　without　considering　IMC(non-IMC)　as　its　total　cost　is　cheaper　according　to　life　cycle　cost　(LCC)　calculations.　In　addition,　failure　probability　based　on　the　proposed　prediction　models　was　also　calculated　as　a　direct　indicator　for　evaluating　the　oil–paper　insulation　status.　©　The　Institution　of　Engineering　and　Technology　2020