To investigate the suitability of TED talks for academic listening in EAP contexts, this research paper compares Academic Vocabulary List (AVL) representation (Gardner & Davies, 2014), lexical density, and speech rate in a TED talk corpus and a lecture discourse corpus, which were both compiled for this study. 28 lecture series (727 lectures total) and 49 TED talks were analysed for AVL representation. TED talks were found to have lower AVL representation than the university lectures (t(75) = 4.95, p < 0.0001). 43 one-minute samples from the Lecture Discourse Corpus and 47 one-minute samples from the TED Talk Corpus were analysed for lexical density, where no differences were found; and speech rate, which was found to be significantly faster in TED talks, in terms of syllables per second (t(98) = 4.23, p < 0.0001) and words per minute (t(98) = 4.20, p < 0.0001). A negative correlation was found between lexical density and syllables per second in the lecture discourse corpus (r = −0.343, p < 0.05), where none was found in the TED talk corpus (r = −0.031, ns), perhaps due to TED talks being a scripted genre. It is concluded that TED talk variation enables a range of academic listening applications.