Purpose – Research has shown that university students, particularly non-native speakers of English, encounter difficulties with various academic text-types and often lack the ability to organise the information in a structure considered effective by the discourse community or to use the significant language features of the text-type effectively to achieve the intended purpose. The study examined the persuasiveness of argument texts written by proficient and less-proficient undergraduates in a Malaysian university. The aspects examined were the organisational structure of the text and the language features significant for fulfilling the persuasive purpose of the text. Method – Forty argument texts were analysed: 22 written by undergraduates who were more profi cient in English, and 18 by less proficient undergraduates. The analysis focused on the organization of the content of the argument text as well as selected language features, namely, connectors, modal verbs and passive voice. Findings – The results showed that the proficient undergraduates used the structure considered effective for argument texts but the writing of the less-proficient group was characterised by unclear or absence of statement of stance and restatement of stance. To achieve the persuasive purpose of the text, the proficient undergraduates made use of connectors, modal verbs and passive voice more than the less-proficient group. However, the connectors ‘because’, ‘so’ and ‘besides’ were often used in a manner similar to spoken language, and there was an overdependence on the modal verbs ‘can’ and ‘will’. The lack of conditionals and nominalisations was obvious for both groups, suggesting that these are more difficult language features to master and greater pedagogical attention is needed. Significance– The paper explores differentiated focus in the teaching of academic writing at university level for students with high and low proficiency in English.