The authors report on the building of an information system for a correspondence school with both many students and many types of students, for which they recommended a fourth-generation-language approach. Because they were sceptical about the massive productivity gains claimed for fourth-generation-language development, and because they did not know exactly where the savings would be, they collected data on the effort expended in the various states of system development. The school needed early schedule and effort estimates to make adequate staff assignments. The authors describe the development approach, discuss the estimation methods used, and present and analyze the results. >