Cognitive architectures enable the modeling of human performance at a detailed level. Data at this level is seldom used to analyze simulated performance to evaluate human-machine systems. To remedy the main obstacles we developed SimTrA (Simulation Trace Analyzer), a method and tool to process simulated and human performance data and to compare the outcomes in an effective and efficient manner. We show the applicability and benefits of SimTrA by presenting the results of a study. We successfully used simulated and human eye movement data to discriminate between two cognitive models and human performance data in a dynamic control task.