A substantial body of data and a number of reviews of the literature on risk encouraged this attempt to construct a theoretically consistent approach to research on risk. Much of the research can be seen as attempting to estimate the probability that development will be compromised for groups of infants born under different circumstances. Linear models have been replaced by the transactional model, which promotes the use of multiple regression techniques to evaluate large data sets involving numerous variables. More needs to be known about the transactional model if fruitful hypotheses are to be formulated about it. Nevertheless, when it comes to developmental outcome in at-risk children, researchers suspect that a great amount of residual variance can be accounted for by transactional dynamics. This paper: (1) explores dynamics and elements in residual variance; (2) suggests an approach to research on risk that promises significant gains in understanding of determinants of normal and abnormal developmental outcomes for at-risk infants; (3) considers the proposition that the concept of risk is useful at each nodal point of developmental change; and (4) advances the notion that the idea of being at risk should encompass not only the risk for downside development, but also the risk that potentially gifted and talented individuals will fail to achieve their developmental potential. Forty-six references are cited. (RH)