This paper discusses the limitations of probabilistic approaches to human error within the context of aircraft certification and continued operational safety. These approaches attempt to estimate frequency of human error over time, with a view to calculating a probability. This type of quantification may be familiar to safety engineers experienced in using such methods to estimate equipment failure rates. In engineering, these estimates are used to evaluate equipment safety, to better understand component lifespan, and to inform decisions regarding maintenance or replacement schedules. However, it is not appropriate to analyze human error in the same way that equipment is evaluated. This paper illustrates why probabilistic approaches to human error are limited, using case studies and examples. The issues raised will be relevant to those involved with human factors in initial aircraft certification or in continued operational safety, or to engineers who require an appreciation of the challenges in evaluating operational flightcrew error in making safety decisions.