The NIMH Research Domain Criteria (RDoC) initiative seeks to utilize multidimensional patterns of socio-cognitive behavior to improve understanding of mental illness. We aimed to evaluate the psychometric properties of a subset of RDoC tasks. Specifically, we investigated two positive valence tasks and five cognitive tasks. Participants (N = 320) were recruited through an online research platform. We used generalizability theory to estimate reliability, and factor analysis to examine factor structure. Reliability was average to excellent with some notable exceptions. Factor analysis results raised concerns about whether the factor structure of task scores aligns with the proposed RDoC model. Effects of cognitive manipulations generally supported the construct representation of tasks. Results indicate that the majority of RDoC task scores examined have acceptable reliability or can be made reliable through modest increases in task length. Future research in diverse populations is needed to better understand the factor structure of RDoC cognitive and positive valence measures.