Although　various　types　of　sensors　are　now　available　to　monitor　a　structure,　measurements　are　usually　obtained　in　only　a　few　locations　numbering　less　than　the　total　degrees　of　freedom　(DOFs)　of　the　structure.　The　lack　of　structural　responses　of　the　structure　at　all　its　critical　areas　may　hamper　the　effectiveness　of　structural　monitoring.　Therefore,　multi-scale　response　reconstruction　at　those　key　structural　locations　where　sensors　are　not　available　is　essential　to　fully　achieving　the　structural　monitoring　objectives.　This　paper　addresses　a　problem　of　placing　multi-type　sensors　in　a　structure　with　the　objective　of　best　reconstruction　of　structural　responses.　The　types　of　sensors　include　accelerometers,　displacement　and　strain　measurement　sensors,　all　of　which　are　widely　used　in　civil　engineering　structures.　The　number　and　locations　of　the　three　types　of　sensors　are　determined　aiming　to　best　possibly　reconstruct　the　strain,　displacement　and　acceleration　responses,　in　which　the　Kalman　filter　algorithm　is　employed.　By　minimizing　the　overall　reconstruction　error　variance　at　the　locations　of　interest　and　maintaining　reconstruction　errors　to　a　desired　target　level,　the　initial　set　of　candidate　sensor　locations　is　reduced　to　a　smaller　set　The　key　multi-scale　structural　responses　are　reconstructed　from　the　fusion　of　limited　multi-type　sensor　data　information.　A　simplysupported　overhanging　steel　beam　is　investigated　as　a　sample　case　in　numerical　and　experimental　study　to　investigate　the　effectiveness　and　accuracy　of　the　presented　approach.　The　good　response　reconstruction　results　clearly　demonstrate　the　effectiveness　of　the　proposed　optimal　placement　method　for　multi-type　sensors.