OBJECTIVE: To assess the prognostic impact of autonomic activity, as reflected by catecholamines and heart rate variability (HRV), in patients with stable angina pectoris. DESIGN: Double blind, randomised treatment with metoprolol or verapamil. 24 hour ambulatory ECG, used for frequency domain analyses of HRV, and symptom limited exercise tests at baseline and after one month of treatment. Catecholamine concentrations were measured in plasma (rest and exercise) and urine. SETTING: Single centre at a university hospital. PATIENTS: 641 patients (449 men) with stable angina pectoris. MAIN OUTCOME MEASURES: Cardiovascular (CV) death, non-fatal myocardial infarction (MI). RESULTS: During follow up (median 40 months) there were 27 CV deaths and 26 MIs. Patients who died of CV causes had lower total power and high (HF), low (LF), and very low (VLF) frequency components of HRV. HRV was not altered in patients who suffered non-fatal MI. Catecholamines did not differ between patients with and those without events. Metoprolol increased HRV. Verapamil decreased noradrenaline (norepinephrine) excretion. Multivariate Cox analyses showed that total power, HF, LF, and VLF independently predicted CV death (also non-sudden death) but not MI. LF:HF ratios and catecholamines were not related to prognosis. Treatment effects on HRV did not influence prognosis. CONCLUSIONS: Low HRV predicted CV death but not non-fatal MI. Neither the LF:HF ratio nor catecholamines carried any prognostic information. Metoprolol and verapamil influenced LF, HF, and catecholamines differently but treatment effects were not related to prognosis.