Although there is growing consensus that parental stress is a risk factor in child development, longitudinal studies of its effects are few. This study tested a sample of 231 mother-child dyads in terms of whether the relations between the global experience of stress in mothers (perceived stress scale) and child temperamental characteristics (infant behavior questionnaire and child behavior questionnaire) could be conceptualized through transactional models of development. The assumption was that higher negative emotionality and lower positive affectivity in the infants would contribute to an increase in maternal stress over a five-year period, beginning in infancy, and that higher maternal stress would contribute to an increase in child negative affectivity and a decrease in positive affectivity and self-regulation over the same period. Evidence was found for both hypotheses, but not within the same models: the effect of maternal stress on child temperamental development was greater. The results are discussed with reference to bidirectional models of temperamental development.