Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia affecting approximately 3 million Americans, and is a prognostic marker for stroke, heart failure and even death. 12-lead electrocardiogram (ECG) is used to monitor normal sinus rhythm (NSR) and also detect AF in ICU and ambulatory patients. Current techniques to discriminate NSR and AF from single lead ECG suffer several limitations in terms of sensitivity and specificity using short time ECG data which distorts ECG and many are not suitable for real-time implementation. There is a clear need for more robust detection and classification algorithms for clinical applications and specifically for delivering appropriate therapy for implantable cardioverter defibrillators (ICD) to provide lifesaving timely action. In this work, the authors propose and demonstrate the application of a multiscale frequency (MSF) technique which takes into account the contribution from various frequencies in ECG and thus yield valuable information regarding the chaotic nature of AF. In this work the authors used AF (25 ECG samples) and NSR (25 ECG samples) traces from publically available Atrial Fibrillation Physionet database for accurate discrimination using MSF approach. The results demonstrate that MSF index is significantly higher (p<0.01) in AF compared to NSR thus enabling robust discrimination. These results offer huge promise for clinical diagnosis of AF from single lead ECG enabling novel treatment strategies in a quick and effective fashion especially in ICD’s as well as for routine monitoring of ambulatory patients. The results also motivate the use of this technique for analysis of other cardiac arrhythmias such as ventricular tachycardia (VT) or ventricular fibrillation (VF).