Too often, ergonomics is relegated to being a “post-design” evaluation, leaving ergonomists little opportunity to make significant and important design changes. One way to start attacking this problem is to study the process of design and, in particular, ergonomics in design. This article describes the findings from a four-month long participant-observer study of the relationship between ergonomics and engineering design. The study was conducted in the context of a large, interdisciplinary project consisting of design of a control room for a nuclear power plant. It was observed that designers and ergonomists must negotiate through a changing web of constraints from many sources. The impact that these constraints had on the course of the design was documented. A model is developed based on the abstraction hierarchy (Rasmussen, 1985, IEEE Trans. Systems Man Cybernet. SMC-15, 234–243; 1990, Int. J. Ind. Ergon. 5, 5–16) which shows the interaction of conflicting goals as ergonomists and other designers attempt to solve a complex design problem. This model leads to several insights: (1) locally optimal ergonomic designs may not be globally optimal, (2) ergonomists can improve their solutions by understanding the goals of other designers, and (3) future tools to aid ergonomists must be compatible with the constraint-rich environments in which they work.