The main objective of this paper is to categorize the values of seem in a parallel corpus of technical texts in English and Spanish. Considering syntactic and pragmatic issues, we offer a taxonomy of seem forms. This also includes the classification of this verb as an evidential or as an epistemic device. Data for analysis were taken from the Corpus of English-Spanish software localization at the University of Vigo (LOGALIZA), which is a compilation of technical texts in English with their corresponding translations into Spanish. The results of our analyses show that the translation of seem largely depends on the translator's own assumptions concerning the meaning of this form in the original texts during the process of translation. We conclude that seem has an evidential function in a high percentage of the analyzed samples.