Two models of recently reported high workload associated with vigilance tasks are the direct-cost and indirect-cost views. The former attributes high workload to the need for continuous observation in discriminating signals from neutral events; the latter attributes it to efforts to combat the boredom associated with monotonous vigilance tasks. These opposing views were tested by providing observers with reliable cueing, which rendered observation necessary only when low-probability critical signals were imminent, or with knowledge of results (KR) regarding performance efficiency. On the basis of cue and KR differences in elicited observation activity and motivational value, the direct-cost model led to the anticipation that cueing would result in a high-boredom, low-workload profile and a greater reduction in workload than KR. The indirect-cost model led to the prediction that cueing would result in a high-boredom, high-workload profile and a lesser reduction in workload than KR. The results clearly supported the direct-cost view that the workload of vigilance is task-induced. Consequently, efforts to combat high workload in complex automated systems requiring substantial monitoring by operators should focus specifically upon task-related determinants.