This paper will describe the experiences of participants in the DO project with the use of structured analysis. The DO project uses the Tevatron II particle accelerator facility at Fermilab. Tevatron II operates within a ring with a four-mile circumference, and is so nameed because it can accelerate particles to energies of a tera (10 to the 12th power) electron volt. The project involves placing a collision detector at a point on the accelerator ring (location DO) where beams of protons and antiprotons can be made to collide. Data from the various instruments within the detector will be captured and analyzed by a complex, multiprocessor real-time system. The software effort must deal in real time with 40 megabytes/second of input data, and must perform detailed analysis on an ''interesting'' subsample of this input amounting to about 2 terabytes per year. The project scope envisions an estimated 100 man-year effort over 3 years, producing 200,000 to 400,000 lines of finished code. One of the major difficulties in the project has been problem definition. The fact that the researchers are both users and implementers has the drawback that one cannot simply refuse to start design until requirements are fully agreed on bymore » all parties - there is no easily-defined administrative division, since many people are involved both in hardware and software development, and deadline pressures occur in various areas for ''finished code.'' Many requirements are in fact fairly nebulous, and one of the uses of formal design has been in specifying what was in fact required.« less