The aim of this paper is to question the relevance of the complement-modifier distinction in Noun-Noun sequences. It will be argued that neither syntactic nor semantic criteria provide a completely reliable basis for the distinction between complement and modifier in the context of post-head complements or modifiers and even less so in the case of nominal complements or modifiers in pre-head position, i.e. in Noun-Noun sequences. More specifically, it will be contended that the distinction between complements and modifiers in Noun-Noun sequences cannot be held on cognitive grounds either. With this aim it will firstly be shown that there are different types of associations between the two nouns in Noun-Noun sequences, namely thematic-relation associations (e.g. food shopping) and peripheral associations (e.g. strip shopping). Secondly, evidence will be provided to show that, in spite of the fact that these various associations may correspond to our intuitions about the complement-modifier distinction, they manifest a similar degree of semantic bondedness and combination frequency. In order to measure the semantic bondedness and frequency of the different types of relations in Noun-Noun sequences a pilot study on a sample of Noun-Noun sequences will be conducted using corpus data. The relative frequency with which two nouns are combined correlates with different degrees of ‘semantic bonding’, which can be seen as a sign of cognitive relevance. As a result of this, it will be shown that, even in the case of relational nouns, the most frequent combinations are not always thematic-relation ones. In addition, some sequences where nouns are combined with other nouns denoting peripheral relations show a higher degree of semantic bonding than others in which they are combined with thematic-relation nouns. Finally, the distinction between complements and modifiers in Noun-Noun sequences will be addressed from the point of view of interpretation. In spite of the fact that thematic relation interpretations are assumed to have priority over property interpretations, this priority may be reversed by different factors.