Assessment of baroreflex sensitivity (BRS) using vasoactive drugs is widely accepted in clinical research. It has been reported that decrease in BRS is an indicative of increased risk of sudden cardiac death. Due to its clinical importance, alternative baroreflex assessment methods have been developed which eliminate use of drugs and since then they are called non-invasive methods. In this study BRS index obtained by drug based assessment was predicted from subset of noninvasive BRS indices extracted from heart rate and systolic pressure time series. A leave one out method was employed to search subset of indices which gives the highest correlation with invasive and predicted BRS. Two predictors provided the highest correlation (0.87). The algorithm selected consistently normalized cross-power in Mayer frequency band and average magnitude square coherence in high frequency band as predictors for all 16 subjects.