English vowels differ in spectral and durational properties. Mandarin learners of English rely mainly on duration to categorize some English vowel contrasts (e.g., /i/-/ɪ/), whereas native listeners predominantly use spectral cues. As Mandarin is a language that does not have vowel duration contrast, Mandarin listeners’ use of duration cues could be attributed to the development of duration categories during the L2 learning process. To examine whether Mandarin learners of English can develop perceptual categories based on duration only, this study investigates their perception of Australian English (AusE) vowel contrasts /ɐ/-/ɐː/ and/ɔ/-/oː/ in both /hVd/ and /hVt/ context. AusE vowels /ɐ/ and /ɐː/ are mainly contrasted in length, and /ɔ/ and /oː/ are contrasted in both length and spectrum. All vowels are longer before a voiced than a voiceless coda consonant. Duration of the four vowels in /hVd/ and/hVt/ context is varied in 11 steps with end points 85ms and 335ms and the interval between steps 25ms. With 4 vowels, 2 contexts, and 11 steps, this yields 88 stimuli. Participants are Mandarin listeners who have studied in Australia for more than 6 months and native AusE listeners. Participants are asked to perform a 2-alternative forced choice perceptual categorization task for each contrast in each condition. The findings are: 1) both groups show categorical perception of duration along the /ɐ/ and /ɐː/ continua, but the category boundary differs between groups in location and steepness; 2) Mandarin listeners are still influenced by duration perceiving /ɔ/ and /oː/ continua, whereas AusE listeners could categorize the stimuli based on spectral features; 3) coda voicing can influence both groups’ category boundary. The results from this study will shed further light on the degree to which Mandarin listeners can and will use vowel duration in the acquisition of English vowels.