Two tests were administered to 87 students at the third through fifth levels of English as a second language (with emphasis on scientific education) at the Center for the Teaching of Foreign Languages of the Autonomous Metropolitan University. Test I consisted of 25 multiple choice and true/false reading comprehension questions based on a scientific text of 400 words. Test II consisted of multiple choice questions based on short texts chosen for the difficulties they presented with regard to vocabulary, rhetorical functions, syntactic complexity, and contextual reference. The scores of the two tests were compared. Although some correlation was found between comprehension and the various points of supposed linguistic difficulty, the correlation was less marked than might have been expected if these points really were causes of difficulty in reading comprehension. The results indicate that a person who understands a text well will also have little or no difficulty with vocabulary, textual rhetoric, sentence complexity, or contextual reference. However, the converse is not true. The study thus fails to find convincing empirical evidence in support of the inclusion of certain types of language proficiency exercises in courses aimed at improving reading comprehension. The tests are appended. (JB)