Given　the　same　alloy　composition　in　Cr5　deposited　surfacing　metal,　the　influence　of　two　different　methods　of　adding　Cr　and　C　in　the　flux-cored　welding　wire　(1#-chromium　carbide　powder;　2#-graphite　and　chromium　powder)　was　studied　on　the　thermal　fatigue　(TF)　performance　of　the　deposited　metal.　Results　show　that　(1)　for　either　as-welded　or　tempered　state,　after　the　same　number　of　thermal　cycles,　the　crack　length　of　specimen　1#　was　greater　than　specimen　2#;　(2)　it　took　about　200　more　cycles　for　the　initiation　of　cracking　and　failure　cracking　on　specimen　2#　than　on　specimen　1#;　and　(3)　after　the　same　number　of　cycles,　the　length　of　failure　cracks　of　the　as-welded　state　was　greater　than　the　tempered　state.　Studies　of　the　microscopic　structure,　number　of　inclusions,　and　grain　size　of　both　types　of　specimens　show　that　only　a　portion　of　chromium　carbide　added　in　1#　flux-cored　wire　was　oxidized　in　high-temperature　molten　droplets　or　weld　pool,　plenty　of　incompletely　decomposed　chromium　carbide　grains　and　those　generated　in　site　lead　to　higher　amount　of　inclusions　in　the　deposited　metal,　compared　with　2#.　In　addition,　the　boundary　between　inclusions　and　matrix　was　distinct,　and　the　contents　of　S　and　P　were　higher　in　1#;　a　larger　proportion　of　residual　austenite　and　twin　structures,　slower　transformation　of　the　residual　austenite　during　tempering,　and　coarser　primitive　austenite　and　matrix　grains　were　found　in　1#.　These　findings　suggest　that　optimization　of　the　method　of　adding　alloy　elements　might　be　an　effective　and　cost-efficient　way　of　improving　the　TF　performance　of　deposited　metal.