A dilemma in designing computer simulations for instruction is how to provide a challenging exploratory environment and yet provide sufficient support so that students do not become lost. Directive support such as corrective feedback may detract from the exploratory quality of a simulation. Two methods of non-directive support are: (1) simplifying the initial presentation of the simulation by having the student manipulate only some of the variables; and (2) presenting the simulation problem to a small group rather than to individuals so that students may give each other feedback on their hypotheses and procedures. This paper examines the effect of both of these methods and their interaction upon undergraduate students solving a problem in a four variable computer simulation of the spread of an epidemic. Subjects were divided into four treatment groups working with varying complexity levels. In terms of instructional effectiveness, as measured by the percentage of students reaching a correct solution, a significantly greater proportion of subjects who worked individually reached the correct conclusion when the task was initially simplified. In contrast, a greater proportion of subjects who worked in groups reached the correct conclusion when presented with full initial complexity. Subjects performed significantly more simulation runs when they were presented fewer variables first and all of the variables later. This study suggests that when a complex, multi-variable simulation is presented, it may be preferable to present it to groups for solution rather than attempt to break it down into component parts. (Contains 17 references.) (Author/AEF)