This research has a four-fold purpose. First, It investigates the linguistic proficiency differences in academic writing between three groups of EFL, ESL, and English L1 master’s students. Second, by employing various statistical methods, it tests and evaluates two sets of lexical and syntactic complexity measures to find out which ones are the best indicators and predictors of linguistic proficiency at the postgraduate level. Third, via a genre-based approach, it seeks out the lexical and syntactic characteristics of dissertation sub-genres. Finally, guided by the principles of corpus linguistics and model-building statistics, this research aims to build a model of lexical and syntactic complexity of sub-genres of academic writing at the postgraduate level.