Classical time and frequency domain analysis was performed on short-term heart rate variability (HRV) signal suitable for clinical trials. The indices derived from both methods were used to study the diagnostic ability of HRV non-invasively. Study considered HRV signals derived from ECG recordings obtained from PhysioNet database. The population under study consisted of 8 normal subjects (control group), 9 arrhythmic patients and 12 patients with severe congestive heart failure (CHF) class III and IV. The derived HRV indices were statistically tested using Receiver Operating Characteristics (ROC). The ability of ROC to discriminate between normal and abnormal patient population was determined by studying the area under the ROC curve.