This thesis reports a study of a corpus of 286 suicide notes collected from the Birmingham Coroner’s Office, with additional findings from 33 real and 33 fabricated notes from Los Angeles. Following some background regarding how suicide notes are treated by Coroners’ Courts and other courts in the U.K, the thesis compares topics used in real and fabricated suicide notes. Although there is considerable overlap between the two categories, they can be partially distinguished by some features that are more likely to occur in one category than the other. For example, dates, indications of author identity and trivia are more likely to occur in real notes than fabricated ones. The thesis then concentrates on fake notes and scrutinises instances of atypical language or phraseology and contextually inappropriate content. It is found that these oddities are far more frequent in the fake notes than in the genuine ones. Finally the thesis focuses on the corpus of genuine notes from the Birmingham Coroner’s Office, using an automatic semantic tagger. The findings are that suicide notes contain significant proportions of items indicating affection, the future and their authors’ kin. In addition, the notes include significant proportions of pronouns, names, negatives, intensifiers, maximum quantity terms, and discourse markers.