The purposes of this study are to examine the effects of the speech-analysis software on Korean college students' pronunciation improvement, and to find out ways to enhance the efficiency of computer-based visual feedback in the teaching of English pronunciation. The participants of the study are 89 English-major college students who are required and expected to have native-like pronunciation. A total of 178 recorded pronunciation data of pre-and post-tests were analyzed in terms of consonants, tense/lax vowels, consonant clusters, and phonological changes. The major findings of the study are as follows: First, the use of speech-analysis software was effective in improving the students' pronunciation. Second, the use of speech-analysis software was more effective at the discourse-level than at the word-in-isolation or sentence levels. In addition, it was more effective to students with low level pronunciation proficiency than those with high level proficiency. Third, the use of speech-analysis software was the most effective in training word-initial [h]-deletion of function words at the sentence level. Fourth, the participants responded that the visual feedback of speech-analysis program was useful in self-monitoring and self-correcting their pronunciation errors. Based on these findings, some pedagogical suggestions are made with respect to the didactic use of speech-analysis software in the teaching of English pronunciation.