Although photon absorptiometry is increasingly used to evaluate bone mineral content of neonates, its reliability in this age group is unclear. The midradius or middistal radius of 12 growing preterm infants was scanned, using a Norland 278 Densitometer. Proper examiner technique and densitometer calibration were first verified. In three infants, the bone edges were not consistently discernible from soft tissue. In the remaining nine infants, apparatus variability (the percent change in bone mineral content of scans repeated without repositioning the infant) was comparable to intraexaminer and interexaminer variability (determined after the same or a different examiner repositioned the infant and repeated the scan). The variability for successive scans ranged from -73% to +60% (SD, 5% to 36%). Adequate reliability for preterm infants remains to be documented for commercially available photon absorptiometers by in vivo studies.