The current study re-examined the “estimation inflation” effect previously found with performance estimates of motor skills. After viewing a demonstration of a balancing task, 54 participants performed either physical practice, imagery practice, or no practice (n = 18 per group). Self-efficacy ratings were obtained regarding perceived ability to perform the task before and after a test phase where actual performance was measured. Prior to the test phase, the imagery group reported significantly higher levels of self-efficacy compared to the physical practice group. However, imagery practice did not benefit performance as only the physical practice group performed better on the balance task compared to control. Thus, imagery practice and physical practice produced dissociable effects on performance estimates and actual performance. Furthermore, inflated performance expectations elicited through imagery disappeared following the test phase. These results provide further evidence that short bouts of imagery can inflate expectations, but not actual performance of a novel motor task. Once individuals gained authentic experience of the task, the initial misjudgement about performance was replaced with more realistic expectations. Therefore, creating overly optimistic expectations of ability did not hamper future expectations once authentic experience was gained.