To facilitate the evaluation of populations at risk for nervous system dysfunction due to environmental agents, we have developed a computer-administered neurobehavioral evaluation system (NES). The system includes a set of testing programs, designed to run on a microcomputer, and questionnaires which are used to record symptoms, obtain exposure history, and characterize potential confounding variables. Standard tasks evaluating memory, visual/motor function, vocabulary ability, and mood were selected and adapted for computer presentation following the recommendation of a World Health Organization (WHO)-National Institute for Occupational Safety and Health (NIOSH) expert committee. After the performance of several pilot studies designed to refine our methodology, validation studies were initiated to assess comparability of selected computer tasks to existing standard ones administered and scored by an interviewer. A different aspect of validity, score stability, was addressed by repeated administration of tests to a small unexposed group. Comparability to existing tests was good (r = .42 to .76). Score stability was excellent over a 150 day interval, particularly for the digit span and continuous performance tests. From our experience, this approach appears to be a feasible, efficient, acceptable, and sensitive approach to evaluating central nervous system function in populations.