Eye-tracking has been used as a methodology for some time in various disciplines, but it has not been applied to date in translation studies. This paper presents results from a preliminary investigation which seeks to answer two questions: Firstly, is eye-tracking, in general, a useful research methodology for investigating translators' interaction with translation memory tools? Secondly, what can eye-tracking data tell us about cognitive load when translators deal with different match types in Translation Memories? The results demonstrate that eye-tracking data, including pupil dilation measurements and gaze replays, in association with retrospective protocols promise to be a very effective methodology for future research into translation processes. The results also suggest that the cognitive load for exact matches in translation memory tools is much lower than for other match types, that cognitive load for machine translation matches is close to fuzzy matches of between 80-90% value, and that 'no matches' require the highest level of cognitive load.