In automated ECG monitoring, QRS detection performance is dependent on noise measurements on individual leads. A new signal quality measurement based on stability analysis of QRS complex features has been developed to assess individual ECG lead quality. The new method was evaluated on the records of the MIT-BIH arrhythmia and NST databases. Results showed that the new signal quality measurement can be used to accurately assess ECG signal quality and can be easily incorporated into an existing multi-lead QRS detection algorithm for performance improvement.