Recent studies have confirmed the importance of automated aircraft data in properly depicting atmospheric structure with data assimilation systems. Typically, automated aircraft data types (i.e., MDCRS and AMDAR) are assigned a low observation error by the data assimilation system, indicating that these data are assumed to be quite accurate, an assumption that has been confirmed by collocation studies. However, these data are not error-free; in fact, there are errors that are quite systematic in nature and so violate the basic assumption that observation errors are random which is made in virtually all data assimilation systems. This paper discusses the nature of errors in currently operational aircraft data, both automated (MDCRS and AMDAR) and manual (AIREP) data types, with examples of some of the more insidious errors presented. These results are a byproduct of the development of the NRL Aircraft Data Quality Control system, a rule-based system that performs both generic error checks as well as checks for specific errors that were uncovered by the generic checks.