The potential of physiologic neuroimaging for contributing to the understanding of behavior and the psychopathologic condition is being enhanced by increased application of "neurobehavioral probes," tasks performed during measurement. Thus far, little attention has been paid to the psychometric properties of such tasks as reliability, difficulty, and construct validity. We propose steps for applying such probes, considering issues in defining the behavior and task selection. Few available neuropsychometric tasks meet criteria for optimal use in neuroimaging studies, and a procedure is outlined for developing new probes. Highlighted are issues encountered during the phases of conceptualization, assembly and screening of items, task construction, and the psychometric validation. A set of language tasks illustrates the process. The procedure may enhance efficiency of acquiring knowledge in this area where the magnitude of potential may be matched by the costs and complexity of research.