1. The accuracy of blood pressure measurement with the Takeda TM-2420 ambulatory blood pressure monitor and the TM-2020 data recorder have been assessed by comparison with simultaneous measurements taken using auscultation and direct femoral artery measurements. 2. Systolic blood pressure was underestimated by the TM-2420 by a mean of 10 mmHg (s.d. = 6, 95% confidence interval (CI) = -13 to -7) over the range of pressures measured by auscultation. It was underestimated by 23 mmHg (s.d. = 12, 95% CI = -28 to -18) compared with direct femoral artery measurements. 3. Diastolic pressure measurements were similar to those obtained by auscultation. When compared with direct femoral artery recordings, diastolic pressure was overestimated by about 5 mmHg (s.d. = 4, 95% CI = 3.4-6.6), which is consistent with indirect readings, taken with a 'standard' cuff (inflatable bladder 23 cm X 12 cm). 4. The TM-2420/2020 is thus suitable for ambulatory measurements of blood pressure when diastolic pressure is the criterion of interest.