Four Vita-Stat (VS) automatic, coin-operated, blood pressure measuring devices were evaluated for accuracy and precision. Under field conditions, 342 adults had two blood pressure measurements with both the VS device and with the Random-Zero (RZ) device. Two of the VS machines gave significantly higher systolic (SBP) and diastolic blood pressure (DBP) values, compared to the RZ values. The mean difference between these two units and the RZ device were clinically important (14.0 and 6.9 mm Hg SBP; 7.5 and 6.8 mm Hg DBP), and resulted in the misclassification of 23% of normotensives as hypertensives. We observed a significant order effect on SBP with the VS device, with a mean decrement of 4.9 mm Hg between the first and second SBP values, compared to 0.6 mm Hg with the RZ device. Even after adjusting for this bias, all four VS devices gave significantly more variable SBP readings than the RZ unit; two of the four also gave more variable DBP values. These data suggest that the VS device is neither accurate nor precise enough at the present time to be recommended for widespread use. These findings also raise questions concerning the monitoring of performance of this and similar devices in the field.