Manufactured　sand　(MS)　was　used　to　partially　replace　the　traditional　quartz　sand　(QS)　to　prepare　a　novel　ultrahigh-performance　manufactured　sand　concrete　(UHPMC).　To　better　understand　the　fracture　damage　mechanism　of　UHPMC,　small-scale　sample　tensile　tests　were　conducted　to　investigate　the　tensile　behavior　of　three　types　of　UHPMC　(low　strain　hardening,　low　strain　softening　and　high　strain　softening).　Acoustic　emission　(AE)　and　digital　image　correlation　(DIC)　are　employed　simultaneously　to　monitor　and　assess　the　damage　characteristics　of　UHPMC.　The　results　indicate　that　the　classical　AE　characteristic　parameters　(ringing　counts,　duration　and　bvalue)　provide　a　satisfactory　result　in　identifying　the　feature　stages　of　low　strain　hardening　and　high　strain　softening.　Based　on　the　Gaussian　mixture　model　(GMM)　and　support　vector　machine　(SVM),　an　automatic　probability　classification　algorithm　has　been　developed　that　achieves　the　quantitative　categorization　of　three　UHPMC　cracking　types　(shear　or　tensile　cracks).　Among　them,　tensile　damage　is　dominant　during　the　entire　loading　process,　whereas　shear　damage,　including　aggregate　dislocation　and　fiber　debonding　and　pullout,　accumulates　during　the　hardening　and　softening　stages.　DIC　visualization　and　AE　source　localization　complement　each　other　in　identifying　crack　distribution,　contributing　to　further　revealing　the　tensile　damage　mechanism　of　UHPMC.　The　results　obtained　in　this　study　can　provide　data　and　theoretical　support　for　fracture　damage　characterization　and　health　monitoring　of　ultra-high-performance　concrete　materials　under　tensile　conditions.