This paper addresses the issue of word-initial rhotic avoidance (WIRA) from a typological point of view. Its first aim is to document WIRA cross-linguistically, based on the examination of a sample of 200 languages designed by the WALS (Dryer and Haspelmath 2013). This set of 200 languages has been surveyed in order to reveal rhotic (and more generally liquid) phonotactic patterns in relation to word-initial avoidance. On the basis of this survey, the paper identifies two types of WIRA: i) phonological, or emic-WIRA; and ii) phonetic, or etic-WIRA. The first and most notable result of this research is that 49% of all languages containing at least one phonemic rhotic exhibit some degree of emic-WIRA, i.e, they possess no word or very few words beginning phonologically with at least one of their rhotics in their native lexicon. The paper also examines how word-initial rhotics are adapted from a non-WIRA language into a WIRA language. The loanword adaptation data suggest that WIRA is a recessive feature because no language in the sample has been observed to develop WIRA due to language contact (although one exception, Gascon, has been identified outside of the 200-language sample). Finally, the paper proposes two new universals in relation to WIRA: 1) if a language forbids /l/ word-initially, it also forbids /r/; 2) a rhotic segment never occurs as the positional allophone of a non-liquid segment word-initially.