The accident at Three Mile Island (TMI) has underscored the need for human factors review and analysis of Nuclear Control Rooms. Virtually every study of the TMI debacle has come to the conclusion that one of the major contributors of the accident was poor human engineering. In recognition of this fact, the nuclear industry, through its research institutes, has initiated a series of research programs designed to provide individual utilities with guidelines on how to apply human factors principles and criteria to improve the safety and performance of their nuclear facilities. The present paper reports on one of the major studies being undertaken by the Electric Power Research Institute (EPRI) in the area. The objective of this research project is to identify and prioritize the existing Human Engineering Discrepancies (HEDs) in nuclear control rooms and to develop retrofits for correcting these HEDs that can be implemented during normal plant operation or during planned outages. Data was collected during one week visits to five power plants. Structured interviews, checklists, task analyses, surveys (noise, light, etc.) and procedural walk-throughs/talk-throughs were employed to assess the design and layout of the control rooms. In depth analyses of these data led to the identification of numerous HEDs. A prioritization scheme was developed to determine the relative seriousness of these HEDs and to rank alternative retrofit solutions in terms of cost and feasibility. The results revealed fairly wide variance in the extent to which control rooms violate human factors principles. The most frequent areas in which violations were found are labeling, alarm/display systems, functional grouping, design conventions, and anthropometric limits. Numerous HEDs along with alternative retrofits are shown and discussed. Emphasis is given to cost and engineering consideration in selecting appropriate backfits.