This study examines the relationship between an open-ended cloze test and its multiple-choice versions generated from three sources: (1) interlingual learner-generated distractors--distractors selected from one language group and administered to a different language group, (2) intralingual learner-generated distractors--distractors selected from one language group and administered to a similar sample of the same language group, and (3) teacher-made distractors--distractors generated by a team of trained ESL/FL instructors and administered to their students. Open-ended cloze tests, multiple-choice cloze formats using intralingual learner-generated distractors, and cloze formats incorporating teacher-made distractors appear somewhat similar in terms of their relationship to general ESL proficiency. In selecting a cloze passage in terms of its discourse, teachers/testers need be aware of and sensitive to their learners' needs and goals. If a language program involves an English for science and technology focus, the appropriate discourse to be taught/tested includes science and technology as academic subjects. If a language program involves a professionally heterogeneous group of learners, a variety of authentic discourse types need to be tested/taught. Some cloze passages are more appropriate than others for language teaching, testing, and learning. Tests and distractors are included. (Author/JK)