The "white coat" effect, an alarm reaction to the presence of a doctor, is an important cause of blood pressure variability, the frequency, amplitude and mechanisms of which are only partially understood. In order to evaluate these factors, a prospective study was undertaken in 35 consecutive patients referred for assessment of clinical hypertension. The alarm reaction was investigated during the consultation, at the time of interrogation, in periods of silence, in the sitting and upright positions. Twenty-four to forty-eight measurements (average 36.8) of the blood pressure and heart rate were performed in each patient with a Diasys 200 R monitor. The ambulatory period of 3 to 5 hours after the consultation provided 12 to 24 measurements (average 20.7) which were considered to be the reference for comparison with the consultation period. A total of 2,038 measurement were made and analysed. Analysis of variance (GLM) for each patient and for the whole group gave an assessment of the alarm reaction during the patient-doctor dialogue and periods of silence with reference to the ambulatory period. During the ambulatory period, the average and standard deviation for systolic pressure were 134 +/- 0.7 mmHg, and for diastolic pressure 93.1 +/- 0.6 mmHg. These pressures were significantly lower than during the two periods of consultation, with and without dialogue (p < 0.0001). During the consultation, the systolic and diastolic blood pressure values were significantly higher during the dialogue than during the periods of silence (p < 0.0001). During the dialogue, the systolic pressure attained 153.7 +/- 0.7 mmHg and the diastolic pressure: 107.2 +/- 0.6 mmHg.(ABSTRACT TRUNCATED AT 250 WORDS)