Human-automation interaction (HAI) takes place in virtually every phase of flight under a variety of operational conditions. Pilots must decide when to engage and disengage automation, which automation to use, the extent to which they should monitor and crosscheck the automation, and so on. Currently, little is known about which characteristics of scenarios involving HAI are important in these decisions and precisely how they influence the interaction. The objective of the present research was to examine how systematic variations of automation characteristics, task features and context variables individually and jointly influence pilots' perception of HAI situations. Commercial pilots received descriptions of aviation situations and were asked to judge the cognitive demands and predict behaviors associated with each HAI event. Results of the study support the notion that HAI comprises a complex interplay between features of the automation and elements of the task and context.