There is interest in how pilots perceive simulator fidelity and rate self-performance in virtual reality flight training. Ten participants were trained to perform a target tracking task in a helicopter flight simulation. After training objective performance, the median tracking error, was compared to subjective self-evaluations in a number of flying conditions where the cues available to our pilots were manipulated in a factorial design: the simulator motion platform could be active or static, audio cues signalling the state of the turbine could be those used during training, non-informative, or an obviously different but informative ‘novel’ sound. We tested participants under hard and easy flying conditions. Upon completion of each test condition, participants completed a 12-statement Likert-scale with items concerning their performance and helicopter simulator fidelity. Objective performance measures show that flight performance improved during training and was affected by audio and motion cues. The subjective data shows that participants reliably self-evaluated their own performance and simulator fidelity. However, there were instances where subjective and objective measures of performance or fidelity did not correlate. For example, although participants rated the ‘novel’ turbine sound as having low fidelity, it behaviourally caused no difference with respect to the turbine sound used in training. They were also unable to self-evaluate outcome of learning. We conclude that whilst subjective measures are a good indicator of self-performance, objective data offers a valuable task-oriented perspective on simulator fidelity.