For routine noncontrast CT examinations of the head, we compared the radiation doses of infant patients aged no more than two years old, with those of "adults" defined as any patient whose weight was greater than 40 kg. Data were obtained for 23 infants, and an equal number of "adults," who underwent CT head examinations between May 1997 and March 1998. Patient CT data acquired included the x-ray tube potential (kVp), mAs, section thickness, and total number of sections. For radiation dosimetry purposes, the head was modeled as a uniform cylinder of water using patient size data obtained from a representative cross-sectional image. CT techniques and patient size data permitted the computation of the mean section doses in the head region, total energy imparted, and the corresponding effective doses. All CT scans were performed at 120 kVp, with an average current-exposure time product of 271 +/- 73 and 340 +/- 0 mAs for infants and "adults," respectively. The radius of the water cylinder used to model the patient head increased from 58 mm for 4 kg newborns to 70 mm for 8 kg infants. For "adults," there was little correlation between the weight of the patient and the mean water equivalent radius of 88 mm (r2 = 0.14). Mean section doses were 44.4 +/- 11.1 mGy for infants, and 44.2 +/- 1.5 mGy for "adults." The energy imparted to infants correlated with patient weight (r2= 0.35) much more than did that of "adults" (r2= 0.02). The average infant energy imparted (66.4 +/- 28.7 mJ) was approximately half the value obtained for "adults" (140 +/- 10 mJ). The average effective dose to the infants (7.6 +/- 3.1 mSv), however, was approximately six times higher than that for "adults" (1.3 +/- 0.1 mSv). There was no significant correlation between patient effective dose and patient mass for either the infant (r2 = 0.12) or the "adult" group of patients (r2= 0.02). Infant doses varied much more than "adult" doses, primarily because of a wider range of x-ray technique factors selected and secondarily due to the variation in infant head size. The observed variability in the computed radiation dose parameters indicates that it should be possible to reduce infant doses routinely in head CT examinations without any adverse effect on diagnostic imaging performance. For such routine head CT scans, the average dose reduction for infants weighing between 4 and 8 kg would be expected to range between 40% and 60%.