Multiple reading measures (i.e., vocabulary, word recognition, sentence comprehension, passage comprehension) and writing measures (i.e., vocabulary diversity, syntactic complexity, qualitative and quantitative measures of spelling and organization) were administered to 256 second graders and 251 fifth graders. An exploratory analysis of these variables was made using canonical correlational analysis. Separate analyses were performed for each grade level sample and for two additional cohorts, beginning and proficient readers, derived from the original samples. The word recognition factors drawn from the reading set were most related to the spelling variables of the writing set at both grade levels and at the beginning reader level. Substantial differences were found across reading level cohorts, however. For proficient readers, the ability to structure prose in complex ways and to use a variety of vocabulary in writing was related to a prose comprehension factor. In none of the analyses was reading or writing found to explain more than 45% of the variance in the opposite test set.Recently, there has been renewed interest in the nature of the relation between learning to read and learning to write. One reason for this upsurge in interest has been a fundamental shift in the theoretical orientations of education and psychology (Anderson, Spiro, & Montague, 1977;Calfee, 1981;Curtis & Glaser, 1981; Piaget & Inhelder, 1969). Reading is no longer described as a passive activity in which subjects, essentially with accuracy, decode and record traces of external reality (i.e., text) in memory. Instead, according to the constructivist notions that currently predominate, readers construct or create messages through a variety of active processes (i.e., schema construction and elaboration, prediction, construction of meaning parameters, information strategies, etc.). Such theories suggest that learners actively use prior knowledge to create information in reading and writing. These explanations are This study is based in part on a doctoral dissertation completed at the College of Education at the University of Delaware. The author wishes to thank