This study proposes a method of cardiac risk assessment based on the long term Holter ECG recordings. The risk assessment is important to prevent the sudden cardiac death incidents which are one of the major cause ofdeath worldwide, e.g. 70,000 in Japan and 400,000 in the U.S. annually. Conventional risk assessment indices are obtained by short term ECG record recorded at the clinical laboratory in the hospital. Such practice tend to miss important symptoms because of the short observation period. For that reason, characterization of the long term ECG record draw a considerable attention. This research adopted several such indices based on the long termECG record for the cardiac risk assessment. Namely, Indices based on QT and RR intervals, such as cRRI-QT, RRI-Amplitude, QT-Amplitude, QTc-Amplitude, SDNN and those based on T wave morphology as AR or ARPare introduced and examined. Logistic regression analysis is applied to those indices obtained from 11 cardiac high risk (SCD-H), 14 low-risk (SCD-L) patients and 25 control subjects (Control). It has been shown that the combination of RRI-amplitude, ARP and cRRI-QT yielded the best classification accuracy. Sensitivity andspecificity were larger than 0.8 except for SCD-L sensitivity being 0.7. The number of cases should be increased to validate the result.Key Words : SCD risk assessment, Holter ECG, T-wave alternans, Heart rate variability, QT-RR intervalco-variability, Logistic regression analysis.