In the literature on Chinese word formation, the (possible) distinction between the processes of ‘derivation’ and ‘compounding’ is still an unresolved issue (see e.g. Pan, Yip & Han 2004; Dong 2005; Arcodia 2012). Word-formation elements which display high productivity and always appear in a fixed position with respect to the base word (in a particular usage), such as 学 xué ‘branch of knowledge’ (as in 心理学 xīnlǐxué ‘psychology’) have been analyzed as affixes (e.g. Yip 2000), as ‘affixoids’ (类词缀 lèicízhuì; Ma 1995) or just as compound constituents (Dong 2004). In this paper, we propose a reassessment of prefixation in Modern Chinese. Following Arcodia (2012), we discard the ‘prefix’ vs. ‘prefixoid’ distinction, since grammaticalized morphemes in Chinese (as well as in most languages of the Mainland East- and Southeast Asian area; Bisang 1996) very often do not show the formal correlates of grammaticalization (i.e. ‘secondary grammaticalization’ in the sense of Traugott 2002). In the framework of Construction Morphology (Booij 2010), we treat potential prefixoids as fixed slots in a construction. In this analysis, the main difference between affixes/affixoids and regular compound constituents lies in their fixed position, their stable selectional properties and, above all, in the fixed, conventionalized meaning they contribute, as opposed to the more ‘open’ interpretation for compound constituents (Scalise, Bisetto & Guevara 2005). The sample items we chose for our analysis are drawn from a selection of the literature on the topic (see the Appendix below); following Basciano & Bareato (2020), we shall rely on web corpora (as e.g. the BCC Corpus and the Leiden Weibo Corpus) for the analysis of the use of complex words. We will show that potential prefixes in Chinese have different properties: there are class-mantaining prefixes (as 前 qián- ‘former’), class-changing prefixes (as 多 duō- ‘multi-’), as well as prefixes with ambiguous properties with respect to word-class assignment (as 非 fēi- ‘non-’). We will compare ‘native’ patterns and patterns which seem to follow a foreign model, showing that they do not constitute coherent subsets in terms of their behaviour. We will argue that the differences between prefixes and suffixes in Chinese (see Jia 2019) may be partly explained by the different role of lefthand constituents and righthand constituents in compounding (unlike e.g. Romance languages). However, as conventionalised constructions used for word formation, prefixation patterns also have properties which do not fit in the general picture of headedness and word-class assignment in the morphology (and syntax) of Modern Chinese: above all, the fact that the word class of ‘prefixed’ words is often inconsistent with that of the corresponding base (non-prefixed) word, as e.g. 贸易 màoyì ‘commerce’ > 非贸易 fēi-màoyì ‘non-commercial’, but both endocentric nouns and adjectives are generally right-headed in Chinese (Ceccagno & Basciano 2007). We will argue that this is a major difference between prefixed and suffixed items in Chinese, since the latter always seem to define the word class of the complex word; also, it can be taken as an argument in favour of analysing prefixes as a separate morphological phenomenon, distinct from suffixation and from compounding.