In this paper, cooperative control is investigated and applied to chained processes with multiple steps and multiple tools in semiconductor manufacturing. A cooperative control architecture is proposed to optimize product quality, to improve yield, to achieve best tool performance, and to minimize throughput time. The architecture consists of three tiers: the top tier for target optimization and overall product performance, the middle tier for tool selection based on tool performance, throughput time and tool availability, and the bottom tier for tool level run-to-run control. Large data sets are collected from four individual process steps in a fabrication facility of a leading semiconductor manufacturer and the data sets are processed and lined up for the study of cooperative control. Monte Carlo simulations are carried out based on the real data to demonstrate a significant improvement for the end-of-line product quality. (C) 2008 Elsevier Ltd. All rights reserved.