The assessment of skeletal integrity by the measurement of ultrasonic velocity through the calcaneus has only recently become widely available and is usually made in conjunction with the measurement of broadband ultrasonic attenuation. Using data obtained with a contact ultrasonic bone analyser (CUBA) system, this report examines whether ultrasonic studies of the heel require the measurement of true velocity of sound in the calcaneus (Vbone), or whether heel velocity (Vheel, defined as the mean velocity through bone and soft tissue) or time of flight velocity (Vtof, defined as the mean velocity between the two transducers) are adequate surrogates. The populations selected for study were 15 healthy young women (group 1, mean age 26 years), 231 healthy peri- and postmenopausal women (group 2, mean age 52 years) and 33 osteoporotic women with confirmed vertebral fracture (group 3, mean age 66 years). Precision was studied by performing 10 repeated scans on the subjects in group 1 and duplicate scans on 144 women randomly selected in groups 2 and 3. Precision was expressed as the percentage coefficient of variation (CV). Both precision studies yielded similar results. The precisions (and 5% to 95% ranges) for all groups combined were: Vbone 2.71% (1465-1809 m/s); Vheel, 1.10% (1511-1646 m/s): Vtof, 0.70% (1349-1425 m/s). Although the precision data suggest Vtof should be preferred, when the range of clinical values is taken into account the smaller CV is exactly cancelled by the narrower range.(ABSTRACT TRUNCATED AT 250 WORDS)