Over 50 colleges witnessed the recent trial implementation of computer-mediated Spoken College English Test (SCET), in which retelling was adopted as one of the new test items. This reform highlights the second attempt to use retelling in the filed of high-stake language tests after the Spoken Test for English Majors—Band Four (STEM-4). To research into the validity of retelling as a test item on the basis of the four aspects of test validity proposed by Lyle Bachman, this paper covers a discussion on the construct validity, content validity, criterion related validity and face validity of retelling. The author has designed a retelling test and gathered a high Pearson linear coefficient (r=0.708) between the term scores and the retelling scores of the two experimental groups, and by comparing the lecturers’ ranks and retelling ranks, an even higher Spearman’s correlation coefficient (rs=0.814) was obtained. These two coefficients are convincing indicators of high concurrent criterion-related validity of retelling. In terms of face validity, the researcher explored the stakeholders’ opinions by distributing questionnaires modeled with 5-Point Likert Scale, revealing a moderate supportive attitude (mean>3.0). The paper has concluded that retelling can be used as an efficient and integrative oral test item for advanced non-English majors.