Several models that attempt to explain stigmatization of people with mental illness emphasize the relevance of social categorization. However, research on illness-based social categories has been limited to explicit self-report measures. In this study, we explored whether implicit social categorization based on mental illness labels can be assessed with the “Who-Said-What” (WSW)-paradigm. In this paradigm, participants have to allocate a large number of statements to individuals who belong to different social groups (e.g., mentally ill vs. healthy). Due to the amount of statements, errors in allocation are to be expected. A disproportionate amount of erroneous allocation to individuals within a social group is interpreted to indicate implicit social categorization. In three studies, we (1) pilot-tested a WSW-paradigm for mental illness categorization (n = 24), (2) replicated these findings in an independent, larger sample (n = 85), and (3) aimed to rule out alternative explanations for the implicit categorization effects with a modified WSW-task and novel stimulus material (n = 137). We found consistent implicit categorization effects with overall medium to large effect sizes (range: 0.41 ≤ d ≤ 1.01). Implicit components of mental illness based social categorization can be assessed with the WSW-paradigm. Future research needs to explore the interrelations of implicit categorization, implicit vs. explicit stigmatization and discriminating behavior.