With the current emphasis on short product delivery lead times and customized product configurations, there is a need for systems that can quickly assemble small batches of customized product. While some plants are converting their assembly lines to assembly cells to achieve this goal, the reasons for the performance improvement resulting from conversion have not been well documented or understood, making it difficult to know when and where assembly cells are applicable. This research adds to the sparse body of literature in this area by examining the planned conversion of an assembly line to a set of parallel assembly cells in a real plant. Analytical and simulation models are used to explain why the proposed cells are expected to outperform the current assembly line.