This paper discusses the role of intelligence in language comprehension. In the study reported, an attempt is made to determine how much of the variability in English reading scores of non-native speakers of English can be predicted on the basis of intelligence alone. In the study, three measures of reading comprehension are used: a traditional multiple-choice test, in which a reading passage is followed by a series of questions; a paraphrase recognition test, which requires that the testee choose from among four choices the sentence closest to the probe; and a cloze test, which requires the testee to fill in the blanks in a passage with words that are either close to or identical to those omitted. It is predicted that of the three tests, the traditional multiple-choice test will be more highly correlated with intelligence than the other two measures, because traditional multiple-choice requires reasoning ability: the ability to retain facts, to draw conclusions, and to use contextual clues to determine the meaning of unknown words. Results allow the conclusion that reading sub-tests found on widely used measures of non-native speaker language ability are tests which, to some degree, measure intelligence, and the validity of the tests is therefore called into question. (Author/CLK)