Errors are an everyday concern in organizations, particularly in hospitals, in which errors are occasionally argued to be a result of workload. In this research we aim to improve our understanding of the relationships between workload and errors. We explore a counterintuitive case of a decrease in error rates when workload is heavy. We drew on sensor technologies used to locate individuals, that provides an opportunity to generate and test alternative potential explanation. Our results are based on data collected by 1000 sensors every three seconds for more than two years of real-time location in 6 oncology infusion units in one ambulatory hospital. We integrated the sensors’ data set with data set of patients’ scheduled appointments, and with data set of error reports. Integration of the three data sets allowed us to show that nurses’ positive adaptive behavior during heavy workload circumstances leads to a valid process improvement that decreased errors in heavy workload situations.