Abstract: Speech intelligibility is an important index for evaluating speech quality in speech halls. Although there are several subjective evaluation methods of speech intelligibility in speech halls, the results of these evaluation methods vary widely. In this paper, three commonly used subjective evaluation methods of speech intelligibility, including 10 subjective evaluation tests, are compared under 15 sound field conditions with four different STI values in classrooms. The results reveal: (1) The evaluation methods present significantly different results under the same STI values, and the worse the listening condition, the more obvious the difference in scores, with the greatest difference reaching 54.1 percentage points. This difference can be attributed to several factors, including the information redundancy of test lists, response mode, the number of test items, types of word and sentence lists. (2) Information redundancy difference greatly affects the subjective evaluation results of monosyllable word lists, disyllable word lists and sentence lists, with the maximum score difference between the three methods reaching 32.0 percentage points, and this effect also varies with listening conditions. (3) The response mode has a great effect on the subjective evaluation results of rhyme word lists, and the worse the listening condition, the greater the effect, but no effect on the matrix sentence lists. (4) The number of test items significantly affects the difficulty and reliability of phoneme balance monosyllable word lists, increasing the number of test items in the test lists leads to lower scores and smaller standard deviations. (5) For the monosyllable word methods, the word list type has a relatively small effect on the subjective evaluation results; for the sentence methods, the sentence list type has a great effect on the subjective evaluation results in the poor listening conditions, and essentially no effect when the listening conditions are good. These findings provide a reference for the subjective evaluation of speech intelligibility in speech halls.