The NIMH-Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) initiative requires, among other things, the establishment of a reliable, valid, and consensus-derived method of assessing cognition. The derived battery will provide a standardized way to assess the effects of cognition-enhancing agents across clinical trials. To this end, the first of six consensus-oriented conferences was held April 2003. The goals were twofold: (a) To select which cognitive constructs to measure in a consensus battery, and (b) to select which criteria to use in evaluating tests for inclusion in the battery. Based on consultation with experts on the RAND Panel Method, 74 experts were invited to participate in a pre-meeting survey to provide information relevant to decisions on the cognitive battery. The survey included sections on reliability, validity, test administration, norms and interpretation of tests, cognitive domains and their integration, battery duration, and overall importance of test qualities. For selection of cognitive targets, the results showed that experts ranked executive functions, attention/vigilance, memory processes, and problem-solving ability highest. For test qualities, the experts ranked test-retest reliability, good coverage of key individual cognitive constructs, and comparable alternate forms highest. This article presents the results of the pre-conference survey that was the first step in the RAND process towards development of the NIMH-MATRICS consensus battery to assess cognition in schizophrenia.