This study examined computerized adaptive testing and cognitive ability testing of adults with cognitive disabilities. Adult subjects (N=250) were given computerized tests on language usage and space relations in one of three administration conditions: paper and pencil, fixed length computer adaptive, and variable length computer adaptive. Subjects were classified into primary disability categories: medical, mentally ill, chemically dependent, brain injury, and no disability. Forty percent of subjects had multiple diagnoses, half of them with both chemical dependency and mental illness. Only three percent were female. Ages ranged from 20 to 76 years. Subjects taking the computerized forms perceived the tests to be easier, faster, more easily read, and more enjoyable than those taking the paper and pencil tests. Test time was shortest under the variable length condition. The mentally ill subjects took longer to complete computerized testing than other subjects. There were no differences in subject satisfaction with the test as a function of ability. Two factors emerged from factor analysis, the first comprising verbal abilities, math and language skills, recent and remote memory, and freedom from distractibility, and the second comprising perceptual abilities, abilities to process nonverbal materials, and psychomotor skills. Findings suggested that subjects performed somewhat better on the computerized version and that clinicians were less accepting of computerized assessment than were patients. (DB)