ABSTRACTReading involves simultaneous top‐down processing, where the reader applies prior knowledge to the text, and bottom‐up processing, where the reader derives meaning from macro‐level structures in the text. Text redundancy is central to both.This study investigated the usefulness of using grammatical cohesion to evaluate comprehension on the macro‐ or discourse‐level, as well as on the micro‐level. The Cohesion Cloze was used to determine the extent to which 65 native speakers of Hebrew made use of text redundancy or cohesion when reading an English text. According to the categorization of grammatical and lexical cohesion by Halliday and Hasan (1976), blanks were chosen to test students’ recognition of the repetition of a word or its pronoun, a synonym, superordinate, or generalization. Students were directed to fill in the blanks with wordsalreadyfound in the text. Items were analyzed in terms of discourse analysis as relations of cohesion and coherence in the text.Results indicated that there was a relation between anaphora and coherence that contributed to reading difficulty. Cohesive markers were helpful where blanks were multi‐determined and there was redundancy of coherence. Passages which focused on a single idea were found easier to fill in than passages containing ideas that were compared and contrasted. Thus shifts in argument. were found to be more difficult to comprehend than a continuous c