The effect of cuff bladder size on blood pressure measurement has been investigated in 838 children aged 5-7 years, using a Dinamap oscillometric automated blood pressure recorder and a Hawksley random zero sphygmomanometer. With both instruments the smallest (infant) cuffs recorded higher pressures than the child cuffs (mean systolic differences: Dinamap 4.6 mmHg, Hawksley 6.6 mmHg), which in turn recorded higher pressures than the largest (adult) cuffs (mean systolic differences: Dinamap 5.5 mmHg, Hawksley 8.1 mmHg). These differences in measurement between cuffs are sufficiently large as to be of concern in both clinical and epidemiological studies. Since they are little affected by arm circumference, their occurrence cannot be prevented simply by following current guidelines for selection of cuff size. Methods of dealing with these problems are discussed.