The lack of specificity in the detection of ventricular tachyarrhythmias remains a major clinical problem in the therapy with ICDs. The stability criterion has been shown to be useful in discriminating ventricular tachyarrhythmias characterized by a small variation in cycle lengths from AF with rapid ventricular response presenting a higher degree of variability of RR intervals. But RR variability decreases with increasing heart rate during AF. Therefore, the aim of the study was to determine if the sensitivity and specificity of the STABILITY algorithm for spontaneous tachyarrhythmias is related to ventricular rate. Forty-two patients who had received an ICD (CPI Ventak Mini I, II, III or Ventak AV) were enrolled in the study. Two hundred ninety-eight episodes of AF with rapid ventricular response and 817 episodes of ventricular tachyarrhythmias were analyzed. Sensitivity and specificity in the detection of ventricular tachyarrhythmias were calculated at different heart rates. When a stability value of 30 ms was programmed the result was a sensitivity of 82.7% and a specificity of 91.4% in the detection of slow ventricular tachyarrhythmias (heart rate < 150 beats/min). When faster ventricular tachyarrhythmias with rates between 150 and 169 beats/min (170-189 beats/min) were analyzed, a stability value of 30 ms provided a sensitivity of 94.5% (94.7%) and a specificity of 76.5% (54.0%). For arrhythmia episodes ≥ 190 beats/min, the same stability value resulted in a sensitivity of 78.2% and a specificity of 41.0%. Even when other stability values were taken into consideration, no acceptable sensitivity/specificity values could be obtained in this subgroup. RR variability decreases with increasing heart rate during AF while RR variability remains almost constant at different cycle lengths during ventricular tachyarrhythmias. Thus, acceptable performance of the STABILITY algorithm appears to be limited to ventricular rate zones < 170 beats/min.