Skill acquisition scientists can rely on measures of bias, precision, and accuracy to describe and capture changes in action outcomes. Here we examined whether bias and precision are dissociable components of an individual’s accuracy. We first explored this by simulating two-dimensional accuracy at two timepoints and varied the persistence (i.e., “stickiness”) of bias and precision. When bias and precision were persistent, Time 1 accuracy was the strongest predictor of Time 2 accuracy. When precision was persistent and bias was not, Time 1 precision was the strongest predictor of Time 2 accuracy. If bias and precision are dissociable, then current accuracy is not always the best predictor of future accuracy. We then re-analyzed data from a previous experiment that likely included the above-described scenarios. The experiment involved participants tossing beanbags to an occluded target in pretest and retention without feedback and 50% relative feedback frequency in acquisition. Bias was random on the pretest due to the blind throw and no feedback. Consistent with simulations where precision was persistent and bias was not, pretest precision was a significantly better predictor of retention accuracy than pretest accuracy. However, after a couple blocks of receiving feedback, accuracy became the better predictor of future performance, consistent with simulations where both precision and bias are persistent. These results suggest bias and precision may be dissociable components of accuracy. Consequently, even if accuracy is the only measure of interest, including pretest precision rather than accuracy as a covariate may increase statistical power in similarly designed experiments.