Three topics are addressed: (1) measurement issues (e.g., the reliability and validity of neurobehavioral test scores), (2) general principles of assessment, including test selection, and (3) interpretation of scores. Psychological tests generally perform as well as medical tests in terms of reliability and validity. Test manuals, assessment textbooks, and psychologists are useful resources to the risk assessor. The variety of different tests employed in neurobehavioral studies complicates interstudy comparisons. In addition, tests that ostensibly assess the same general domain of function might assess somewhat different abilities within that domain. Although a uniform battery for use in all studies seems desirable, the battery appropriate for a specific study depends on study goals, knowledge about the mechanism(s) of neurotoxicity, nature of the study population, and pattern of exposure. Exposure-related neurobehavioral deficits are generally indicators of "altered function" rather than of "clinical disease." Limiting concern to end points corresponding to clinical disease might not be appropriate. Many neurobehavioral diagnoses are phenomenological and a neurotoxicant might cause a unique pattern of deficits for which no label has been created. The concern that a small shift in the central tendency of a distribution of test scores has no significance for the individual should be reexamined in light of the prevention paradox, formulated on the basis of epidemiologic studies of chronic disease. Poor performance on a neurobehavioral test does not necessarily map clearly onto underlying behavioral or neural substrate. The absence of such linkages, given current knowledge about brain-behavior relationships, should not reduce confidence in neurobehavioral end points. Use of neurobehavioral test scores involves considerations that differ little from those that the risk assessor routinely addresses in using end points commonly used in research on other topics in environmental epidemiology.