An　approach　for　estimating　the　non-line　changing　of　the　leakage　power　and　the　speed　of　the　dual　threshold　domino　OR　gates　with　increasing　fan-in　based　on　Wavelet　Neural　Networks　is　proposed.　Sine　all　of　the　average　estimating　errors　are　less　than　5%,　the　estimating　system　has　high　precision.　The　reasons　for　the　estimating　errors　are　the　changing　ratio　of　the　different　power　in　the　total　power　and　the　effect　of　capacitor　match.　At　last,　through　Monte　Carlo　analysis,　the　availability　of　the　estimating　system　under　process　variation　is　tested　and　concludes:　With　process　variation,　the　estimating　system　is　applicable　to　leakage　power　estimation　with　low　fin-in　and　delay　estimation　with　high　fin-in.