The psychometric characteristics of the Graduate Record Examinations General Test (GRE-GT) were studied for three handicapped groups. Experimental subjects took the GRE-GT between October 1981 and June 1984; they include: (1) 151 visually-impaired students taking large-type, extended-time administrations; (2) 188 visually-impaired students taking timed, regular-type administrations; (3) 108 physically-handicapped students taking timed, regular-type administrations; and (4) 441,654 non-handicapped students. Data on a non-handicapped subset (N=20,499) who took the GRE-GT in October 1981 and April 1982 were included. Psychometric characteristics studied included the level of test performance, test completion, test reliability, and unexpected differential item performance. Results indicate that the performance of visually-handicapped groups closely approximates that of non-disabled students. However, physically-handicapped students performed substantially lower than did the comparison group of non-handicapped students on two of the GRE-GT scales. Data on test completion showed evidence of possible disadvantage to the two groups of disabled examinees taking timed administrations. Test reliability analysis via internal consistency and parallel-forms methods indicated no practical differences in measurement precision across groups. In terms of both reliability and item behavior, the psychometric characteristics of the GRE-GT appear comparable for handicapped and non-handicapped groups. Nineteen data tables and 11 sample GRE-GT questions are provided. An annotated bibliography of five reports of the "Studies of Admissions Testing and Handicapped People" is appended. (TJH)