Time-variant　reliability　evaluation　including　finite　element　(FE)　analysis　requires　relatively　large　computational　efforts,　as　FE　analysis　is　time-consuming　and　the　number　of　calls　to　FE　analysis　in　time-variant　reliability　analysis　(TRA)　cycle　is　generally　large.　In　this　study,　a　moment-based　PHI2　(MPHI2)　method　is　proposed　for　evaluating　time-variant　reliability　including　FE　analysis,　based　on　the　statistical　moments　of　the　limit　state　function　at　successive　time　instants.　Instead　of　conducting　FE　analysis　repeatedly　in　TRA　cycle,　MPHI2　method　separates　FE　analysis　from　the　cycle　by　representing　the　associated　components　in　the　limit　state　function　as　a　random　variable/process　and　estimating　its　statistical　characteristics　before　the　TRA　cycle.　After　the　separation,　time-variant　reliability　can　be　evaluated　based　on　the　assumed　random　variable/process　with　no　need　to　conduct　FE　analysis,　which　may　significantly　reduce　the　number　of　calls　to　the　FE　analysis　and　thus　improve　the　computational　efficiency.　Three　numerical　and　practical　examples　are　presented　to　illustrate　the　application　and　efficiency　of　the　proposed　MPHI2　method.