In air traffic control (ATC), forgetting to perform deferred actions—prospective memory (PM) errors—can have severe consequences. PM demands can also interfere with ongoing tasks (costs). We examined the extent to which PM errors and costs were reduced in simulated ATC by providing extended practice, or by providing external aids combined with extended practice, or by providing external aids combined with instructions that removed perceived memory requirements. Participants accepted/handed-off aircraft and detected conflicts. For the PM task, participants were required to substitute alternative actions for routine actions when accepting aircraft. In Experiment 1, when no aids were provided, PM errors and costs were not reduced by practice. When aids were provided, costs observed early in practice were eliminated with practice, but residual PM errors remained. Experiment 2 provided more limited practice with aids, but instructions that did not frame the PM task as a “memory” task led to high PM accuracy without costs. Attention-allocation policies that participants set based on expected PM demands were modified as individuals were increasingly exposed to reliable aids, or were given instructions that removed perceived memory requirements. These findings have implications for the design of aids for individuals who monitor multi-item dynamic displays.