A project is described that was undertaken to investigate: (1) how long a cloze test has to be to achieve optimum reliability without wasting anyone's time; and (2) how cloze tests should be scored in order to obtain maximum reliability. The literature recommended 50 deletions in order to provide for an adequate sample of examinee's abilities; it did not strongly recommend either exact-word scoring or acceptable-word scoring. A 50-deletion regular cloze test and a multiple-choice cloze test were administered, respectively, to 50 and 57 foreign students. The regular test answer sheets were scored by the exact-word, acceptable-word, and clozentropy methods. It is concluded from examination of the data that the acceptable-word method is superior to the other two methods and to the multiple-choice cloze test. It was also shown that by 25 items, the maximum reliability had almost been achieved; at the same time it was clear that the best ratio is obtained through tests with 50 items. It is recommended that the most efficient use of everyone's resources can be achieved in a cloze test of 25 deletions and use of the acceptable-word method of scoring. (AMH)