This　paper　introduces　a　lifting　double-wavelet　algorithm　for　electrocardiogram　signal　denoising　by　using　the　improved　threshold　function　in　the　lifting　wavelet　decomposition　to　deal　with　wavelet　coefficients.　The　results,　based　on　data　from　MIT-BIT　arrhythmia　database,　show　that　noise　can　be　eliminated　effectively　with　lower　distortion　and　higher　signal-to-noise　ratio　compared　with　the　single　wavelet　thresholding　method.　The　algorithm　is　also　highly　suitable　in　real-time　signal　processing　systems　and　wearable　medical　devices.　©　2016　IEEE.