This article provides preliminary validity evidence for the shorter Mandarin version of the Vocabulary Size Test (VST) under the content aspect, technical quality, substantive and generalizability aspect of Messick's (1995) construct validity framework. The shorter version with 177 Chinese university students in three proficiency levels indicates that the test enjoys a high level of validity both from the perspective of classical test theory and item response theory. A one-way ANOVA and detailed statistics of Rasch analyses demonstrate that the test could effectively distinguish learners from different proficiency levels. The first eight levels of the VST generally display an ascending difficulty continuum except that some loan words in the seventh and the eighth level prove to be problematic in assessing Chinese learners' vocabulary size accurately. The generalizability aspect of construct validity is demonstrated in that there is no item bias for learners of the same proficiency group in terms of gender. Compared with the results of other studies either on monolingual or bilingual version, this research attempts to justify the appropriateness of the Mandarin version and calls for cautious applications of the test instrument on Chinese learners.