The present study investigated whether self-estimation bias in maths competence was associated with motivational costs or benefits, concurrently and prospectively, and whether these associations depended on students' achievement levels. Participants were 2,807 seventh grade students who were reassessed after 2 years to examine the longitudinal effects on motivation and achievement attributable to overestimation of their own competence. It was found that high-achieving students experienced short-term as well as long-term benefits with overestimation, although motivational benefits flattened out and no further benefit was observed after an optimal level of overestimation. However, low-achieving students experienced short-term motivational benefits even with extreme overestimation, but these short-term benefits resulting from extreme overestimation turned out to be a 'poisoned chalice' without the skills and knowledge to bolster the inflated self-confidence.