In configuration problems, such as the construction of a weekly study schedule, decision makers must assemble a combination of parts under a set of constraints. Interactions may be present between the parts, and more than a single objective function may exist, such as minimizing the number of days on campus and maximizing the interest level of the courses. Little is known about the decision-making processes involved in configuration tasks. Building on problem-solving and resource-allocation literature, the current study examines the effect of the starting points on performance in these tasks. We predicted that task constraints serve as starting points for the construction process that pull the entire configuration toward them. In 2 experiments, we asked participants to construct a weekly course schedule. We manipulated the course-offering sets so that starting the configuration with one of the task constraints (mandatory courses or a requirement to be on campus on specific days) and myopically building the configuration around it would either help or hinder the effort to reach a good schedule. By several convergent measures, participants succeeded more with the "Asset" course-offering sets than with the "Obstacle" sets. Think-aloud protocol analysis also supported this starting-point pull effect. The results of Experiment 3 show that for a starting point to pull the configuration, it has to be a constraint on the task. All 3 experiments show how building the configuration myopically around the starting points unduly affects the end result. (Contains 3 tables, 1 figure, and 8 footnotes.)