Abstracts in technical writing are a genre that functions more than a mere summary situated at the beginning of academic report writing. They are considered a very important site that promotes and draws the readers' attention to read on or to disregard the paper. Often, writers tend to use evaluative language in abstracts to convey their stance in disciplinary knowledge making and to convince the readers of the value of their study. This study investigates the authorial stance and engagement with the readers in postgraduate theses abstracts. It consists of a corpus of 44 MPhil and PhD online theses abstracts by Chinese students from the departments of Electronic Engineering, Computer Engineering and Information Technology, and Manufacturing Engineering and Engineering Management in a Hong Kong university. The analysis of texts is based on the four parameters of evaluation in texts proposed by Thomson and Hunston (2000), namely: the good/bad or positive/negative parameter; the certainty parameter; the importance parameter; and the expectedness parameter. The findings show that evaluative language signaling the worth of the study and the writer's certainty of the knowledge claim and the good-bad parameter were mostly used in the corpus. This study also discusses its implications on technical communication pedagogy.