Human Error Data. Facts or Fiction?

The need for human error data for various purposes has been dis-cussed for decades, yet no acceptable human error data bank has emerged. What is the problem? Are there events which can objectively be considered human errors and for which data can be collected from real-life work situations? What attributes are necessary to characterize the human involvement in accidental chains of events? In the paper, these questions are discussed, and it is argued that instead of focusing on human errors, data should be collected to represent situations of human-task mismatch and characterized accordingly. Furthermore, to support design of error-tolerant work situations, more emphasis should be put on analysis of error recovery features