This study examined the oral reading miscues of eighteen subjects, three average readers each from grades two, four, six, and eight, plus three high and three low readers from grade four. The miscues produced by each student were analyzed using the Goodman Taxonomy of Reading Miscues. Length of text was most important for low readers, who produced more totally acceptable miscues as they progressed through the text. All of the groups had a strong control of syntactic structure and all were better able to control syntactic structure than meaning. All of the groups relied somewhat upon preceding material when attacking meaning. Further, these groups demonstrated that readers test their guesses in regard to what syntactic structure and what meaning is coming next. When their guesses prove inconsistent, either syntactically or semantically, readers often attempt to correct. A relationship can be made between the use of context in reading varying portions of text and methods and materials for the testing of reading. (Author)