The aim of this paper is to present a prosodic description of ideophones in Teko, contributing to the recent development of research on ideophones (Childs 1994, Voeltz and Kilian-Hatz 2001, Dingemanse 2012). The prosody of ideophones, in particular, is rather little investigated (Dingemanse et al. 2016, Reiter 2011). Teko (aka Emérillon) is a Tupi-Guarani language spoken by 400 persons in French Guiana (Glottocode emer1243), and has been recently described (Rose 2011). A lexical database of 1327 lexemes (Rose 2018) contains 103 ideophones, “marked words that depict sensory imagery” (Dingemanse 2011:25). The corpus of 38 texts shows 388 occurrences of ideophones on a total of 2000 sentences. The phonological, morphological, syntactic, semantic and discourse-pragmatic features of Teko ideophones have been previously described by Rose (2011:400-409). They set aside ideophones as a distinct part of speech, found in both narratives and conversation to depict situations (actions or states), with different degrees of syntactic integration: ideophones can either be an optional addition to a clause, be introduced by a ‘do/say’ verb, or be used as the unique predicate of the clause. The paper investigates various aspects of the prosody of ideophones. First, we will show that a syllabic pattern ending with a final closed syllable (such as in moŋ ‘dark’ and dɨɾɨg ‘look’) is statistically favoured in ideophones (66%) in comparison with other parts-of-speech (45%), χ2(1) = 15.61, p < .001. Second, we will highlight the fact that two thirds of Teko ideophones show vowel harmony across the entire word, a process that is not usual in the rest of the lexicon (see also Akita et al. 2013). Third, acoustic measures in Praat will confirm the perception of ideophones’s prosodic saliency due to vowel lengthening, high pitch and intensity, as opposed to “plain” prosody (Gordon and Rose 2006). Studying the prosodic variation of the same ideophone in different contexts also helps explain how prosody can add semantic information to the lexeme, for example vowel lengthening and a salient pitch drop in the ideophone tou ‘fall’ indicates that the object falls from a very high location. Fourth, the length of pauses around ideophones will be evaluated, as was done for direct speech (Rose and Vanhove 2007). The third and fourth tasks are based on one hour and forty minutes of audio-recorded texts, and enable the exploration of the relation between prosodic and syntactic integration, following the methodologies of Malibert and Vanhove (2015) and Dingemanse and Akita (2017). To summarize, this paper will first complement the existing description of Teko ideophones, and second contribute to the emerging typology of ideophones, especially their prosody, a promising under-investigated domain for ideophones. It provides answers to three general questions: Is the prosody of ideophones in general distinct from that of other parts-of-speech? How does the prosody of ideophones participate in their semantic interpretation in context? Does the prosodic integration of ideophones match their syntactic integration?