The major purposes of the study reported on in this paper were as follows: (1) to compare the errors made by advanced learners of English as a native language (American nine- to eleven-year-olds) and as a foreign language (native Spanish-speaking adults); (2) to investigate the usefulness of a modified cloze test in obtaining errors for profitable analysis; and (3) to aid teachers and researchers in evaluating the potential of error analysis as a tool by providing a demonstration of what it can reveal about language learning. Given a test consisting of a series of highly-contextualized cloze-type items, the subjects were asked to supply a word for each blank which would satisfactorily complete the sentence. Those responses judged by adult native English speakers to be errors were classified by the experimenter into six (non-exclusive) categories, entitled context alteration, non sequiturs, grammatical errors, lexical confusions, uninterpretable, and other (more minor). The individual error categories and the pattern of error contributions by the two subject groups are discussed, although the errors contributed to the categories by the two groups showed an overall similarity both in terms of quantity and quality. (Author)