This　paper　studies　the　sampled-data　cooperative　output　regulation　problem　for　linear　multi-agent　systems.　Firstly,　a　novel　adaptive　distributed　continuous-discrete　observer　is　established　to　recover　the　leader's　state,　which,　on　one　hand,　only　relies　on　the　sampling　output　of　the　leader,　and　on　the　other　hand,　does　not　need　to　store　the　sampled　message　from　neighbors.　Secondly,　by　solely　making　use　of　the　digital　states　of　both　the　agent　and　the　distributed　observer,　a　certainty　equivalence　control　law　is　synthesized　featuring　a　time-varying　feedforward　gain.　It　is　rigorously　proven　that,　with　this　time-varying　feedforward　gain,　the　proposed　control　approach　can　achieve　exponential　convergence　of　the　tracking　errors　given　arbitrary　time-varying　leader's　signal,　and　the　upper　bound　for　the　sampling　intervals　is　explicitly　given.　Thirdly,　for　the　class　of　chain-integrator　multi-agent　systems,　the　proposed　control　approach　does　not　need　any　restriction　on　the　upper　bound　of　the　sampling　intervals,　and　thus　would　be　more　practical　in　certain　application　scenarios　from　the　perspective　of　energy　saving.　The　performance　of　the　proposed　control　approach　is　validated　by　the　simulation　results　of　inverter-based　distributed　generation　systems.　IEEE