OBJECTIVE This study aimed to identify and characterize cognitive subtypes among youth at clinical high risk (CHR) for psychosis. METHODS We conducted a cluster analysis on cognitive measures in a large sample of CHR patients (CHR; n = 764) and healthy controls (HCs; n = 280) from NAPLS-2 (North American Prodrome Longitudinal Study 2). These findings were independently validated using a comparably large sample from NAPLS-3 (n = 628 CHR, 84 HCs). Statistical approaches were employed to compare cognitive, clinical, and functional trajectories at baseline and during 24 months of follow-up, with further analysis of conversion status within these clusters. RESULTS Two cognitive clusters, "impaired" and "intact," were identified in the CHR group. Baseline differences in verbal abilities and attention and working memory domains distinguished the cognitively intact cluster from HCs. Longitudinally, the impaired group displayed a 'floor' effect, with no noticeable deterioration, but showed a "catch-up" trajectory in attention and working memory. This group had a higher conversion rate and more cases diagnosed with psychotic disorders than the intact group. In the intact group, converters exhibited a decline in attention and functioning. Most cognitive trajectories showed a positive relationship with functional outcomes. CONCLUSIONS The study provides evidence for distinct cognitive subtypes in CHR, independent of conversion status. Early evaluation across cognitive domains is crucial for identifying trajectories to tailor interventions that aim to enhance outcomes for individuals at CHR for psychosis. Future research should focus on longer follow-up trajectories targeting attention and working memory and explore related intervention strategies.