In linguistics a corpus typically involves a finite body of texts which are considered to be representative of a particular variety of language at a specific time (McEnery and Wilson, 2001). Those are the assumptions we have had in mind in this metaphor corpus based on business press headlines. Our body of texts is a finite number of headlines drawn from the specific field of the business sections of three newspapers:' Financial Times', 'El Pais' and 'El Mundo', published over a period running from January to July 2003. Compiling a small corpus of non-literal instantiations as different authors have done (Cortes de los Rios, 2001; Kövecses, 2002; Charteris-Black, 2003; Koller, 2004; Deignan, 2005; and others) will enable us first to identify whether the contextual meaning of a word or a multiword unit of headline contrasts with its basic meaning and whether the contextual meaning can be understood by comparison with that basic meaning, and then to categorize, both in the Spanish and in the British press, the different linguistic realizations of a headline in terms of their syntactic structure, metaphor foci and source domains. (Honesto Herrera Soler)