To supervise automated production systems, operators have to allocate their visual attention among a variety of information sources to maintain their understanding of the current system state. In the present study, we investigated how this process is affected by the interplay of different task characteristics (task demands, time on task, and task reward) and their effects on dimensions of the operator’s mental state (mental workload and mental fatigue). To this end, 50 participants performed a process monitoring task with ten experimental blocks, which alternated between low and high task demands. In addition, the last two blocks included a task reward manipulation. We measured eye fixation durations and positions as indicators of attention allocation. In line with our hypotheses, the analyses revealed that participants’ fixation durations decreased with higher task de-mands, increased over time on task, and decreased with the addition of task re-ward. We also found that participants’ attentional focus narrowed over time on task and widened again with the addition of task reward. Our results demonstrate how operators can cope with changing task demands by adapting their attention allocation. The results also highlight safety implications of mental fatigue, as both slower shifts in operators’ visual attention and restricted attentional focus create the potential for operators to miss critical changes in system state.