Capillary refill has been advocated as an indicator of perfusion status (shock) in seriously ill patients. An upper limit of normal of two seconds has been recommended; there is no published evidence that supports this value. To investigate the validity of the two-second upper limit of normal and to examine the variation of capillary refill with age and temperature, we measured capillary refill in 100 healthy child, 104 adult, and 100 elderly volunteers. In addition, 20 adults were measured before and after a one-minute immersion in 14 C water. Median capillary refill times for the young female, young male, and adult male volunteers were 0.7, 0.8, and 1.0 seconds, respectively. These times were significantly shorter than those of the adult women, elderly women, and elderly men groups, whose median times were 1.2, 1.5, and 1.8 seconds, respectively. In the temperature experiment, preimmersion times were significantly shorter (median, 1.3 seconds) than those after immersion (median, 2.9 seconds) (P less than .01). We conclude that capillary refill is age and temperature dependent. Application of the two-second upper limit of normal to our populations would result in a false-positive rate of 4.0% for the pediatric and adult male volunteers, 13.7% for the adult female volunteers, and 29.0% for the elderly volunteers. The upper limit of normal for adult women should be changed to 2.9 seconds, and the upper limit of normal for the elderly should be changed to 4.5 seconds if 95% of all normal patients are to be contained within the normal range. The temperature dependence of capillary refill raises questions regarding its reliability in the prehospital setting.(ABSTRACT TRUNCATED AT 250 WORDS)