The　small　punch　test　and　uniaxial　tensile　test　of　P91　steel　were　conducted　at　600　degrees　C,　and　the　reliability　of　finite　element　method　with　the　Gurson-Tvergaard-Needleman　(GTN)　damage　model　was　verified　by　comparing　the　simulation　results　with　test　findings.　The　effect　of　small-scale　specimen　thickness　on　material　properties　was　investigated　through　seven　different　specimens　of　varying　thickness　from　0.1　mm　to　0.7　mm,　using　the　GTN　damage　model.　Finally,　the　traditional　empirical　expression　was　improved　to　correlate　the　small　punch　test　characteristic　loads　(Fy　and　Fm)　and　the　material　mechanical　properties　parameters　(sigma　YS　and　sigma　UTS)　at　high　temperatures.　The　results　showed　that　the　maximum　loads　of　different　specimens　with　varying　thickness　significantly　differ　from　the　0.5-mm　specimen　thickness　specified　in　the　standard　small　punch　test,　and　the　results　of　different　specimens　with　varying　thickness　can　be　correlated　to　a　0.5-mm　specimen　thickness　with　little　modification.　The　material　properties　predicted　by　the　improved　equations　are　consistent　with　experimental　results　except　for　sigma　UTSwith　a　0.1-mm　thickness.