A major source of registration failure in microlithography was found to be due to variations in optical field sizes defined by the dies, caused by unsatisfactory focus control. Two methods for determining variations in optical field sizes are described. Both allow measurements of selected registration errors with an uncertainty of ±0.01 μm (1σ) under manufacturing conditions using commercially available measuring microscopes. The long-term registration stability of stepped-mask exposure systems was also investigated. It is concluded that maintenance of registration over long periods of time can be improved through accurate focus control along the optical axis. Finally, it is suggested that no single set of measurements at one point in time can completely characterize a microlithographic system. Periodic monitoring of key measurable parameters during use is advisable.