Consistent with the belief that passion is beneficial for job performance, prior research has identified positive associations between passion and predictors of performance, including self-efficacy, engagement, and commitment. However, studies that directly examined the passion-performance link have found inconsistent results. To reconcile these disparate findings, we suggest that passion’s high levels of investment and identification cause an outcome detrimental to performance: overconfidence. We test this passion overconfidence pathway by exploring whether passion predicts higher self-ratings (i.e., overconfident performance assessments) above and beyond objective/other-ratings. An initial meta-analytic review (k=28, N=5,951) found evidence consistent with our hypothesis that passion increases overconfidence by comparing self-rated versus other-rated and objective measures of performance. Four subsequent studies with MBA students and employees (N=2,389) compared self-ratings to other-ratings (including supervisors’ ratings) found that passion predicted higher self-ratings but not other-ratings. In a fifth sample, we recruited employee-supervisor pairs from 110 different organizations, and found that passionate employees exhibit overconfidence in part through increased investment and identification. Two experimental studies (one pre-registered, N=509) provide causal evidence that passion increases overconfidence and replicate the mediation pathways of investment and identification. The current research reveals that overconfidence appears to be a poisonous yet predictable side effect of passion, one that can explain why passion inconsistently predicts performance.