This paper analyses the presentation of scientific facts from a cognitive linguistic perspective, where metonymy plays a role in mapping numerical values to appropriate scientific language in medical discourse in English and Croatian. Scientific facts are presented in terms of both their epistemic meaning and their social meaning relevant in the specific discursive community. Scientific facts are commonly expressed as numerical values which represent a certain point on the scale of a reference value. Scientific language reflects the degree or intensity of scientific facts. The role of metonymy as a cognitive process is to generalize and map the vertical, scientific dimension of thinking to appropriate scientific language, thus giving a social component to scientific facts. Different linguistic elements are used to describe relationships on the scale of numerical values and create a factual state in the sense of socially acceptable scientific facts. The relationships on the scale of numerical reference values in English are conceptualized in the domain of [verticality]. Consequently, their linguistic realization in English scientific discourse is achieved by attributes such as high, low, increasing, decreasing as a reflection of the conceptual metonymy up for more/ down for less, leading to the metonymic mapping vertical scale for numerical value. In Croatian, scientific facts are conceptualized in the domain of [quantity] and based on the metaphor more is up/less is down. Thus the adjectives veliki or mali, which are concepts describing quantity, are normally used. However, the influence of English scientific discourse causes changes in Croatian conceptual structures, leading to the use of attributes from the domain of [verticality], which are the result of metonymic mapping. It is known from the cognitive linguistic literature that language is metaphoric in nature and that conceptualizations are embodied and based on human experience. In scientific discourse, there are conventions and systems that underlie the conceptualization of scientific facts. Thus it seems that scientific language is more metonymic in nature.