I appreciated the analysis by Rice and Anderson of the effectiveness of child restraint systems.1 Tennessee was the first state to require the use of child restraint devices, doing so in 1978, and an analysis of their effectiveness in preventing death and preventing or reducing injury was published in 1984.2 That study demonstrated restraint devices to be associated with an 11-fold reduction in risk of death, whereas children traveling in the arms of an adult were exposed to risk of injury or death comparable to that of children left entirely unrestrained (a result that led to the removal of the “babe in arms” exemption from the child restraint law). The reduced relative benefit of child restraint systems (approximately 3-fold) in this most recent study, compared with the first such study (11-fold), might simply reflect variability among studies, but might also reflect 25 years' improvement in other crash-risk factors, including softening of automotive interiors, improvements in highway barriers and other engineering, widespread implementation of emergency medical systems, and so on.