Drawing from an overview of cloze research, this paper sets forth the thesis that the validity of the cloze technique and its usefulness as a measuring device have been clearly established, but its usefulness as a teaching device has not been so determined. Salient points of research are summarized as follows: (1) if rates of developing understanding vary with the student and with the connective, (2) if gains in understanding continue beyond age 12, (3) if types of errors reflect varying levels of understanding, and (4) if understanding connectives is important to reading comprehension, it follows that systematic instruction in context clues, structure words, and other aspects of syntactic relationships should yield increases in reading comprehension scores, regardless of the measures used. A scheme of eight types of context clues is then established, followed by various techniques to translate theory into practice. On the premise that context determines meaning, the paper next shows how an efficient means of transferring learning would be to develop new retrieval systems for the information the student already has. The paper views using context clues as hypothesis-testing and argues that the global communication skills are teachable and learnable through direct instruction by the cloze technique in any content area classroom. The paper concludes with recommendations for instructional applications at any grade level and a note about future research. (FL)