This study examines the effectiveness of integrating commercial pronunciation software into an ESL pronunciation class of learners at university level. The study partially replicates Seferoğlu's (2005) research design and seeks to confirm her findings through a revised methodology. Participants in this study were 18 international graduate students from various departments of the Iowa State University in the US. Students were assigned to two experimental groups which received traditional classroom pronunciation instruction and instruction that integrated the use of commercial pronunciation software, respectively, for six weeks. A pretest and a posttest using the same picture-description task were conducted in an attempt to find changes in the students overall pronunciation quality. As measured by the ratings of comprehensibility and accentedness from six native speaker raters, the group receiving software-integrated instruction did not show significant pronunciation improvement after the treatment. Neither did the two groups show significant difference in their pretest and posttest scores. Therefore the results did not confirm Seferoğlu's findings. In addition, the students' reflection on the instruction received was analyzed to explore which features of the two types of pronunciation instruction were considered most useful and least useful by the language learners.