We begin with the part-oriented definition of a workcell task in terms of generic elementary actions to be performed. These actions are then mapped to a workcell-element-oriented description of the operations to be executed to perform the task. When the workcell task is repetitive, the notion of a workcell cycle becomes important, and we can associate with each workcell element a repetitive sequence of operations. In this paper, we shall demonstrate that workcell programming is more than just a matter of sequencing, since there may exist alternative cycles inherent in the same specification of the workcell tasks. These alternatives are then compared in terms of cycle time, task throughput, task-processing rate, and machine utilization. But even for simple workcell tasks, the space of possible time evolutions of workcell behaviour can be very large, with many different workcell cycles. To automate such sequence analysis, we therefore present a Prolog-based decision-support system called SAGE (sequence analysis by generalized enumeration). Three simple examples are used to illustrate our approach.