In risk analysis and risk research, the concept of risk is often understood in quantitative terms. For example, risk is commonly defined as the probability of an unwanted event, or as the mathematical product of the probability and magnitude of an unwanted event. Such quantitative conceptions of risk are not only common in technical approaches to risk analysis. Also in social science approaches to risk, e.g. literature on governance, regulation and management, risk often is considered having a quantitative nature. This presentation describes a study which addresses (1) to what extent and (2) in what ways that the word “risk” is actually used in a quantitative manner. Uses of the noun “risk” are analysed in four different linguistic corpora – both Swedish and English (mostly American English). A corpus is a database of naturally occurring language use that has been collected for the purpose of linguistic study. The corpora explored in this study are: the Corpus of Contemporary American English (COCA), the Swedish Language Bank (SLB), the Spoken British National Corpus 2014, and finally a corpus of the web sites of seven Swedish Government agencies. In total, more than 16 000 uses of “risk” are studied in 14 random or complete samples from sub-sections of these corpora (e.g. news texts and magazines in COCA and SLB). In contrast to the widespread theoretical assumption that risk is a quantity, the word “risk” is mostly used in a non-quantitative sense. Quantification of “risk” is most frequent in the sub-corpora of magazines, newspapers and academic journals, while least frequent in sub-corpora of government web sites, fiction and spoken language. When used in a quantitative manner, this use is very rarely numerical, but rather rely on less precise expressions of quantification such as “high risk” and “increased risk”. The relatively low frequency of quantification of risk in government discourse is surprising given assumptions of the risk governance literature. Implications of this discrepancy between theoretical focus of risk as quantity and actual linguistic practice are also discussed.