The relationship between anaphoric nominal substitution and reading comprehension was studied. The Diagnostic Reading Test and the Substitution Test were administered to 80 college juniors, seniors, and graduate students in teacher certification courses, and to 92 college freshmen seeking assistance in improving their reading skills. Positive and significant relationships were found between total substitution scores and reading scores for both samples. It was found that anaphoric resolution was a highly complex cognitive-language skill and a factor in reading comprehension, explaining approximately 59% of the variance in reading performance. The process of antecedent-substitute resolution was not a simple matter of recovering the information from the context. The distance a reader had to backtrack to locate the antecedent and the amount of information represented by the substitute were found to be good predictors of difficulty in performing the referencing task. Of the four item characteristics investigated, the syntactic variables change and discrepancy were not as powerful as the semantic variables length and distance in explaining item difficulty. Since readability measures have not incorporated anaphoric relations as a factor in their assessment, it was recommended that a framework for examining the comprehensibility of text would be a valid alternative to readability formulas when it would be necessary to evaluate and select materials for specific readers. (RL)