Plasma spray is typically characterized by large variations and drift over the lifetime of the electrodes, between different electrode sets, as well as from part-to-part.  The impact for plasma spraying thermal barrier coatings is that many parts need additional processing, even for large windows of coating thickness (+/- 20%).  In order to reduce process variation, a number of different control strategies have been proposed.   This paper investigates the performance of a number of alternative control strategies in terms of achieving the primary objective of maintaining desired coating thickness, as well as the resulting coating structure.  Both model-based feed-forward and real-time feedback control strategies are evaluated, as well as choice of different control objectives such as temperature, centroid position, and velocity.  Important implications of torch nonlinearities such as bifurcations as well as sensor issues for control system design are addressed.