Abstract: A longstanding puzzle for regulatory theory and practice has been the valuation of children's lives, or more precisely, the valuation of mortality risks faced by children. There is data about parents' willingness to pay (WTP) to reduce such risks, and that data might be used to produce a Valuation of Statistical Lives (VSL) for children. The problem is that any such VSL comes from the parents' valuations; it might not adequately capture the welfare effects of risk reduction for children themselves. It can be shown, however, that use of parental WTP, and the resulting VSL, is justified if (1) patents have adequate information, (2) parents do not suffer from a relevant behavioral bias, (3) parents are motivated to care about their children's welfare, so that parental judgments about how to allocate limited resources for their children promote their children's welfare; and (4) parents have a limited budget for expenditures on their children, and regulation would amount to a forced exchange, producing a dollar-for-dollar reduction from that budget. If we relax one or more of these assumptions, the VSL for children might be higher than the VSL that emerges from use of parental WTP. If we relax (1), (2), and (3), the analysis of how to proceed is straightforward, at least in principle. If we relax (4), it is probably best not to tinker with VSL, even if the assessment of welfare and distributional effects is not straightforward.