A review of current theory and research related to writing ability leads to the conclusion that there is no existing test, instrument, or set of procedures which will provide valid data regarding the writing abilities of individuals. As an alternative approach, a textual cognition model of written communicative competence is proposed and elaborated. Arising out of the Trent Valley Project of the Ontario Institute for the Study of Education, the model is operationalized through an extension of the cloze procedure from reading to writing. The underlying assumptions are that written communicative competence implies that a student's writing will make sense for an intended or relevant audience and that, in turn, the cloze procedure will provide a valid measure of the extent to which the writing does make sense to that audience. On the basis of preliminary investigations, a taxonomy of impediments to successful written communication is developed for problems at the graphic, syntactic, lexical, and contextual levels. Questions remaining to be answered about the textual cognition model and the associated cloze procedure are listed. (AA)