Using a multidimensional parental power framework, power bases derived from adolescent descriptions of parent-child interactions and family decision making were regressed on a multidimensional power construct and showed differential importance of predictor variables. Rodman's (1972) modified 'theory of resources in cultural context' is supported. Father's education alone does not lead to loss of paternal power; paternal power retention or loss must be considered in relation to such variables as maternal age and education and paternal age, and within relevant power categories. Mother's education affects her power bases; she increasingly shares culturally prescribed maternal and paternal power as her education increases. Her decreases in traditionally prescribed maternal power are not offset by an equivalent increase in traditionally prescribed paternal power; she may actually lose power with increased education. Relative education of husband and wife explains power differentials across power bases more clearly than does education of either spouse.