Subject-field classification systems are implemented in every major termbank to facilitate the internal management of terminographic work, and this study begins with an account of how the DANTERM subject-field classification system was selected to meet such needs in the bilingual Irish-English termbank focal.ie. Little is known, however, about how subject-field labels and other metadata are actually used by users of termbanks. The current study thus also sets out to investigate users’ beliefs and opinions about how the presentation of metadata, and especially subject-field labels, affects user behaviour and success in the context of English-Irish translation, including use of the bilingual Irish-English termbank focal.ie. Users’ opinions and beliefs are investigated in a series of five focus groups involving nineteen users of focal.ie. Actual use of the termbank is subsequently observed in a contextual inquiry, involving observation of and interviews with nine professional translators in their workplaces. Findings from the focus groups suggest that users believe that ‘no information is too much information’ and they prefer collapsible items to be revealed rather than concealed by default. Although all metadata is considered useful, metadata items that put terms into context (examples of usage, related terms) are considered most useful. The contextual inquiry interviews and observation on the other hand suggest that translators work at speeds that mean that they cannot scrutinise metadata in great detail, but that their use of terminological resources is guided by the overarching theme of trust. Translators’ trust in the data and metadata, in the source of the metadata (creator of the metadata), in their client/employer and in themselves had more of a bearing on how they used the metadata than did the amount or the presentation of the metadata.