Agents are intelligent software objects capable of reasoning about, and interacting with, their environment. This paper presents conceptual and prototypical models of an adaptive production control system based on agent technology. In the system presented both workpieces and resources are represented by agents. Supervisory agents are introduced as the production load on the resources increases and removed as the load decreases. These supervisory agents serve to balance the production load across resources and maximize overall throughput. Simulation studies are presented demonstrating improvements in throughput, resource utilization, and work-in-process inventories over systems invariant to production load.