The First ISLSCP Field Experiment (FIFE) was conducted during 1987 on a 15 X 15 km area in central Kansas, including the Konza Prairie Natural Research Area. The experiment goal of understanding and modeling the effects of vegetation on the heat an mass fluxes in the atmosphere/land surface boundary layer required a flexible and responsive information system with georeferenced data to permit the necessary coordinated analyses. The FIFE Information Syseem (FIS) met this goal in a variety of ways, depending on the type of data and the type of scientific investigation involved. The FIS has served the FIFE investigatiors as a tool for experiment design, as well as for organizing and manipulating the complex data set. Two specific requirements were rapid and simple access to specialized user specified data sets and quick evolution as the science needs changed. As a result, the FIS differs in several important ways from "classical" data systems, covering the spectrum of conceptual, managerial, and organizational issues. Our experience indcates that the design, development, and operation of an information system supporting such an experiment must be flexible and under direct day-to-day control of scientist/users. This experience has also led to a number of observations (both technological and organizational) for designing and building both future experiment support data bases and long term data systems supporting scientific research. Key words: FIFE, Information Systems, ISLSCP