A method of predicting an arrhythmia, such as ventricular tachycardia, for example, in a medical device using a quantitative measure in order to allow assessment of patient risk and to enable preventative interventions by the device and clinicians. The trending of day and night average heart rates, along with patient physical activity can be analyzed to provide prediction of impending arrhythmia within weeks. By examining day and night average heart rate for crossover points, where the night heart rate equals or exceeds the day rate, and monitoring for a concomitant elevation in the night heart rate from a reference value, specific days heralding an increased risk of arrhythmia can be determined and therapy can be updated accordingly.