A task analysis was first initiated in this study for the supervisor in process control systems. An experiment was then conducted by simulating a Reactor Water Cleanup System and the effects of some variables were focused upon, e.g. format types (graphical vs. digital), volumes of data (number of figures) and layout methods of data (process, function, importance) on the human-machine interface design. The main effects of format types, volumes, and layout methods of data and the two-way interaction effects, between format types and volumes of data and between volumes of data and layout methods of data, were observed from the results of the experiment to be significant for the searching tasks. Some main effects and the interaction effects were also observed to be significant for the detection and prediction tasks. The importance of the layout method was then suggested here to be preferred for supervisory tasks. Additionally, the optimal volumes of data were roughly eight figures per screen and the graphical formats were better than digital formats for either searching or detection /prediction tasks. Finally, some further researches are recommended.