Previously research lead to software requirements completeness criteria for process control systems. These criteria, however, did not include the human-computer interface. This research involves extending these criteria to specify appropriate abstractions and completeness requirements for man-in-the-loop systems and devising hazard analysis techniques for the human-computer interface. Four questions are examined: (1) What information needs to be specified in a software requirements specification to ensure safety for software that includes a human-computer interface; (2) How can traditional hazard analyses be extended to include human-computer interface issues; (3) How can the information from these hazard analyses be used to improve the human-computer interface design; (4) How can a state-machine-based requirements specification be analyzed for compliance with some of these human-interface criteria, how can specification languages be designed to enhance the ability to perform this analysis, and to enhance the ability to generate safe human-computer interface designs. ***
