. Blood pressure recording In children. Blood pressure levels for 148 children aged 0-14 years were recorded. Two different methods of measurement (the standard sphygmomanometer and the random-zero machine) and two cuffs of varying size were used on each child. Measurements of arm size and pulse rate were taken, and an assessment was made of the emotional state of each child: no association was found between these factors and the recorded measurement of blood pressure. The experiment does however show that: (1) the effect of cuff size can be minimized by using the largest cuff that will fit the child, and then applying a standard correction to the measurement taken; and (2) the random-zero machine is a preferable method of measurement as it reduces observer bias, though measurements taken with a standard and a random-zero machine are not directly comparable.The many factors that appear to affect the validity of indirect blood pressure readings in both children and adults have now been the subject of investigation for more than 30 years. In studies of direct and indirect readings of blood pressure in infants and older children, Woodbury, Robinow, and Hamilton (1938) and Robinow et al. (1939) showed that, when arbitrarily selected cuffs were used, the accuracy of indirect readings was greatly affected by the relation between arm size (as determined by arm circumference) and the size of the blood pressure cuff. They concluded that the smaller the arm, the narrower the cuff should be, and that, for the newborn, cuffs wider than 2-5 cm would produce artificially low readings. In a study of 51 adults, Ragan and Bordley (1941) attributed differences between direct and indirect blood pressure measurements to the influence of arm circumference measurements. It should, however, be noted that some of their subjects were patients with aortic incompetence. The authors also compared the relative accuracy of the standard American 13 cm cuff with the 20 cm cuff, finding that the wider cuff gave more 'accurate' readings where the arm circumference was large, though its use in 'average' subjects produced artificially low readings.