The value of feedback about errors when learning a novel computer-based task was explored in two studies. The first study examined the optimal level of information to be provided in feedback about errors. The second study examined whether framing errors positively as opportunities to learn (which encourages error tolerance) or negatively as hindrances to learning (which encourages error avoidance) facilitated learning and performance. Both studies used a computer-based simulation of a management decision-making task. In the first study there were three feedback conditions: outcome feedback alone, outcome feedback plus error signal feedback, and outcome feedback plus corrective feedback. Corrective feedback produced better performance than error signal and outcome feedback but learning did not differ across the three conditions. Corrective feedback also facilitated the use of systematic exploration which was positively associated with performance and learning. Learners' self-efficacy moderated the effects of error feedback: learners with high self-efficacy showed high levels of performance in all conditions but for those with low self-efficacy, detailed corrective feedback was essential for learning. The second study explored the effects of positive vs. negative error framing and corrective vs. signal error feedback in a 2 × 2 design. Positive error framing produced more unsystematic exploration and worse performance than negative error framing. Positive error framing helped those with low self-efficacy but for those with higher self-efficacy it was of more value to frame errors negatively. The implications of the interactions between error framing, error feedback and learner characteristics are discussed along with implications for the study of error management, a positive error framing technique.