Although the validity of internet gaming disorder (IGD) in the DSM-5 hinges on a relationship between 5 or more IGD symptoms and “clinically significant impairment and/or distress”, to date most studies have focused on statistical significance. To address this, we conduct an individual participant meta-analysis comprised of primary data from 15 studies (n = 38,851). Study 1 finds that meeting the DSM-5’s proposed 5/9 diagnostic threshold is associated with d = .65 greater distress across 21 well-being constructs, which exceeds a conservative anchor for clinical significance. However, we also find that classifying participants above and below the 5/9 threshold has little power to reject that threshold: a 2/9 cut-off predicts similarly large differences. Study 2 shows that dimensional (continuous) modelling of IGD scores offers a more severe test of the proposed threshold. Finally, study 3 reveals that three criteria—preoccupation, tolerance, and loss of control—are roughly half as predictive of distress as withdrawal and escapism, highlighting limitations also present in dimensional modelling and suggesting the need to modify or remove these. In sum, we do not find evidence for invalidating IGD as proposed, but do identify issues with threshold-based categorization, inadequate differentiation between statistical and clinical significance, and the inclusion of potentially flawed criteria. We argue that these are possible negative consequences of a premature switch to confirmatory research on IGD.