Heart　is　one　of　the　most　complex　dynamic　system.　Indeed,　ECG　is　important　information　which　reflect　the　active　states　of　heart.　In　this　paper,　select　24　data　files　of　ECG　from　MIT-BIH　database　and　divide　these　data　into　four　groups.　Series　of　R-peaks　and　RR-　intervals　extracted　from　these　electrocardiograms　are　analyzed　with　the　correlation　dimension　and　correlation　entropy.　The　results　showed　that　there　are　obviously　different　dynamic　characteristics　in　the　four　groups.　Use　the　methods　of　nonlinear　dynamics　to　extract　dynamic　characteristics　from　ECG　so　as　to　research　active　states　of　heart　and　help　doctors　clinical　application　for　early　diagnose　diseases　of　heart　in　future.