This paper describes the application of a target-oriented contrastive analysis model to an extensive corpus of medical research articles. The analysis focuses on the Methods section and a subset of lexical items representing persons viewed as the object of clinical study. Quantitative contrastive analysis revealed statistically significant differences between the translations from English and the independently created Spanish texts in all the thematic, syntactic and lexical variables analysed. Qualitative contextual analysis showed that four basic criteria for thematic position and a series of associated translation strategies are capable of correcting the excesses and deficits observed, thus producing a more natural and acceptable target language text.