This paper argues that Merrill's instructional transaction theory provides a framework that can decrease learners' problems in using computer simulations by specifying transaction shells rather than frames in designing interactive courseware. The three types of data used in a transaction, i.e., an instructional algorithm, are identified as a knowledge base, a resource database, and a set of instructional parameters. Each of these types of data is described in detail: (1) the knowledge base consists of three types of frames (entities, activities, and processes), and a number of elaborations of these frames (attributes, components, abstractions, and associations); (2) the resource database contains mediated representations of the domain knowledge together with presentation techniques to display the results of the model, acceptation techniques to make interventions in the model possible, and communication techniques to transfer data from the model to the educational representation and back; and (3) instructional parameters can be designed for a particular population, learning task, and environmental situation. The value of instructional transaction theory for computer simulations is specified as the definition of domain knowledge to include in the simulation and the separation of the instructional parameters. Examples of instructional parameters are given, and it is noted that further research and development are needed to identify the instructional parameters and the domain knowledge to include in a "process" transaction shell. (Contains 8 references.) (ALF)