Users interact with technical systems based on their mental model of the system. Both, the mental model itself and the situation inhere uncertainty, because mental models are only a reduced representation of the reality and complex systems do not provide a complete set of information. However, users are able to develop strategies to cope with uncertainty and finally, to reach satisfying results. In this study we investigated the development of mental models during the performance of an uncertain task via eye movements. Main results showed decreasing visual search activity parallel to an increasing learning level and hence decreasing subjective uncertainty. In addition, eye movement parameters seem to be able to differentiate between high and low performers in the way that low performers showed more visual search activity. These findings may be of relevance for human-machine interaction by using eye movement patterns to detect individual learning problems and the degree of the user’s subjective uncertainty. Based on the eye movement analysis, the information content of the technical system could be adapted to the user’s learning curve to improve the interaction even in the context of uncertain systems and tasks. However, further studies are necessary to proof the stability of the concept.