Most recent thinking about thevulnerability of research subjects uses a``subpopulation'' focus. So conceived, theproblem is to work out special standards forprisoners, pregnant women, the mentally ill,children, and similar groups. In contrast, an``analytical'' approach would identifycharacteristics that are criteria forvulnerability. Using these criteria, one couldsupport a judgment that certain individuals arevulnerable and identify needed accommodationsif they are to serve as research subjects.Seven such characteristics can be evident inchildren: they commonly lack the capacity tomake mature decisions; they are subject to theauthority of others; they (and their parents)may be deferential in ways that can maskunderlying dissent; their rights and interestsmay be socially undervalued; they may haveacute medical conditions requiring immediatedecisions not consistent with informed consent;they may have serious medical conditions thatcannot be effectively treated; and they (andtheir parents) may lack important sociallydistributed goods. Each of thesevulnerabilities can call for special care inthe design and implementation of researchprotocols.