OBJECTIVE--To determine the long term effects of relaxation therapy on 24 hour ambulatory intra-arterial blood pressure in patients with mild untreated and uncomplicated hypertension. DESIGN--Four week screening period followed by randomisation to receive either relaxation therapy or non-specific counselling for one year. Ambulatory intra-arterial blood pressure was measured before and after treatment. SETTING--Outpatient clinic in Amsterdam's university hospital. SUBJECTS--35 Subjects aged 20-60 who were being treated by general practitioners for hypertension but were referred to take part in the study. At three consecutive screening visits all subjects had a diastolic blood pressure without treatment of 95-110 mm Hg. Subjects were excluded if they had damaged target organs, secondary hypertension, diabetes mellitus, a cholesterol concentration greater than 8 mmol/l, or a history of malignant hypertension. INTERVENTIONS--The group allocated to relaxation therapy was trained for eight weeks (one hour a week) in muscle relaxation, yoga exercises, and stress management and continued exercising twice daily for one year with monthly visits to the clinic. The control group had the same attendance schedule but had no training and were requested just to sit and relax twice a day. All subjects were asked not to change their diet or physical activity. MAIN OUTCOME MEASURE--Changes in ambulatory intra-arterial blood pressure after one year of relaxation therapy or non-specific counselling. RESULTS--Mean urinary sodium excretion, serum concentration of cholesterol, and body weight did not change in either group. Diastolic pressures measured by sphygmomanometry were 2 and 3 mm Hg lower in subjects in the relaxation group and control group respectively at the one year follow up compared with initial readings. The mean diastolic ambulatory intra-arterial pressure during the daytime had not changed after one year in either group, but small treatment effects could not be excluded: the mean change for the relaxation group was -1 mm Hg (95% confidence interval -6 to 3.9 mm Hg) and for the control group -0.4 mm Hg (-5.3 to 4.6 mm Hg). Mean ambulatory pressure in the evening also had not changed over the year, and in both groups nighttime pressure was 5 mm Hg higher. The variability in blood pressure was the same at both measurements. CONCLUSIONS--Relaxation therapy was an ineffective method of lowering 24 hour blood pressure, being no more beneficial than non-specific advice, support, and reassurance--themselves ineffective as a treatment for hypertension.