The retrospective think-aloud method, in which participants work in silence and verbalize their thoughts afterwards while watching a recording of their performance, is often used for the evaluation of websites. However, participants may not always be able to recall what they thought, when they only see few visual cues that help them remembering their task execution process. In our study we complemented the recording of the performance with a gaze trail of the participant" eye movements, in order to elicit more verbalizations. A comparison was made between the traditional retrospective think-aloud protocols and the variant with eye movements. Contrary to our expectations, no differences were found between the two conditions on numbers of problems, the ways these problems were detected, and types of problems. Two possible explanations for this result are that eye movements might be rather confronting and distracting for participants, and the rather generic way of probing we used. The added value might be stronger when specific questions are asked, based on the observed eye movements. Implications for usability practitioners are discussed in the conclusions of this paper.