Eight nurses were re-trained in the technique of measuring blood pressure firstly with an audiotape and secondly with a video and expert instruction. They were then assessed for accuracy in measurement in subjects with a range of BP. In the first session with the audiotape, criteria for accuracy were that the difference in readings (systolic and diastolic) between each pair of observers was to be 5 mm Hg or less in at least 85% and 10 mm Hg or less in at least 95% of the readings. Five out of the 28 observer pairs (18%) fulfilled these criteria for both systolic and diastolic pressures, six (21%) satisfied the criteria for systolic pressure and nine (32%) for diastolic pressure with eighteen pairs (64%) failing to meet the criteria for both systolic and diastolic pressure. In the second training session seven of the original eight observers underwent further training in which the audio-tape was replaced by the British Hypertension Society video and instruction was also given by an expert. The same requirements for accuracy were used except that, in addition, the difference between the trainee observer and the expert had to be within 5 mm Hg in at least 90% and within 10 mm Hg in at least 95% of the readings. Eighteen of the 21 observer pairs (86%) achieved the accuracy criteria, with the remaining three (14%) failing on diastolic pressure only. All seven observers were within the required criteria when assessed against the expert observer. It is concluded that training of observers is more effective with a video and expert instruction than by using audio-tapes.(ABSTRACT TRUNCATED AT 250 WORDS)