OBJECTIVE: To examine the correlates of therapeutic response of patients with panic disorder presenting with palpitations, we hypothesized that therapeutic response would correlate with heart rate variability (HRV) and sleep measures. METHODS: After a 1-week placebo washout, 27 patients free of structural heart disease and not on cardioactive drugs were randomized in a double-blinded fashion to 4 weeks of treatment with clonazepam (a known antipanic agent) or placebo. We performed standard sleep measures and recorded HRV from 24-hour Holter acquisitions at baseline and end of study. We defined response to therapy as a 50% improvement in the Hamilton Anxiety Rating Scale (HARS) score, confirmed by questionnaires and reaction to sodium lactate infusion. RESULTS: There were 12 responders and 15 nonresponders. Normalization of sleep pattern (including less stage 1 and rapid eye movement [REM] sleep) was observed in both drug and placebo responders (P = 0.011 and P = 0.05, respectively) and in placebo responders alone, compared with nonresponders (P = 0.006 and P = 0.013, respectively). Placebo responders were more likely to show less depression, but even after we controlled for depression, main sleep effects remained. None of the HRV measures correlated with response, but compared with placebo, clonazepam led to a decrease in all the time and frequency domain measures of HRV (all P < 0.05). CONCLUSIONS: Central mechanisms are related to the therapeutic response of patients with panic disorder presenting with palpitations, but this does not directly correlate with HRV. Larger and longer studies may allow objective explanations of placebo response in panic disorder.