The present paper addresses the analysis of evaluative stance in newspaper opinion articles with a twofold objective: from a theoretical standpoint, we present a new model for the analysis of evaluative stance in discourse. The model contemplates the following categories: (a) Function: classifying, predicational and attitudinal evaluative categories, each of which is prototypically related to a Part of Speech; (b) Figurativity: metaphoric and non-metaphoric evaluative expressions, and (c) Value: positive and negative connotations. From an applied standpoint, the model is used in order to analyse the frequency and role of markers of evaluative stance in a corpus of British newspaper articles. The corpus consists of two sets of opinion articles from the British newspapers The Times and The Guardian (30,000 words each). Our research questions are the following: (1) What is the frequency of lexical units expressing evaluative stance in the corpus? (2) What is the distribution of the evaluative stance categories across the corpus? (3) Which category or categories express evaluation in a more prototypical way? (4) What is the more frequent value of metaphoric expressions? (5) Are there any significant differences in the expression of evaluative stance in the two newspapers?