Electromagnetic　(EM)　data　obtained　by　using　an　m-sequence　source　is　typically　segmented　with　an　m-sequence　cycle,　and　the　random　noise　is　suppressed　by　stacking　the　earth　impulse　responses　recovered　from　each　data　segment.　However,　the　stack　cannot　suppress　the　powerline　noise,　which　is　normally　several　orders　of　magnitude　greater　than　the　random　noise.　In　this　paper,　the　powerline　noise　reduction　with　different　data　lengths　were　compared　by　numerical　simulation;　the　results　show　that　there　is　always　an　optimal　data　length,　not　a　single　cycle　but　several　cycles　that　can　achieve　maximum　powerline　noise　suppression.　The　concept　of　an　identification　system　was　proposed　and　the　identified　earth　impulse　response　is　obtained　by　measuring　the　system　output　with　the　input　of　the　voltage　response.　The　optimal　data　length　was　summarized　in　two　cases　by　analyzing　the　identification　system　with　different　data　lengths.　Coding　frequency　optimization　was　proposed　to　shorten　the　optimal　data　length　to　increase　the　stacking　times.　Finally,　the　validity　of　the　theory　described　in　this　paper　was　verified　by　field　data　processing.　Different　from　the　traditional　subtraction　method　and　notch　filter,　the　powerline　suppression　method　presented　here　does　not　need　a　noise　record　or　filter　design;　the　powerline　noise　is　suppressed　accompanied　by　the　identification　processing　without　any　additional　workload;　the　simple　and　practical　method　provides　a　new　insight　into　the　powerline　noise　reduction.　With　this　method,　the　utilization　of　field　data　is　improved　and　the　anti-noise　performance　of　the　m-sequence　is　also　fully　exploited.　(C)　2017　Elsevier　B.V.　All　rights　reserved.