The　malfunction　of　measuring　apparatus　is　almost　inevitable　in　electric　system　and　will　result　in　inaccurate　settlement　amount　of　electricity　sales.　The　backup　measures,　such　as　dual-apparatus,　calculations　from　the　measurement　on　other　side　of　line　or　Tele　Meter　Reading　of　EMS　are　helpful　to　rectify　the　erroneous　data.　It　has　to　detect　the　anomaly　according　to　the　inherent　character　of　load　as　those　back-up　absent.　A　method　of　determination　and　correction　of　measurement　anomaly　based　wavelet-threshold　is　proposed　in　this　paper.　It　decomposes　the　measurement　data　in　multi-scale　space　with　wavelet　transform　and　determines　the　scale-coefficient　which　includes　the　anomaly　data　based　on　the　difference　between　normal　load　fluctuation　and　extraordinary　error.　A　modified　threshold　function　is　used　to　separate　the　error　component　from　the　scale-coefficient.　The　correction　component　can　be　reconstructed　from　the　error　component　in　the　binary　tree　of　wavelet　transform.　An　example　is　given　to　test　and　verify　the　method　finally.　©　2011　IEEE.