The verbal item types of the Graduate Record Examinations (GRE) were studied to explore possible reasons for any differences found in subgroup performance and validity. Statistical differences among item types in active forms of the GRE verbal measure were documented; experimental subtests of matching statistical characteristics for the item types were developed; and correlations of the matched and operational item type scores with self-reported grade point averages were compared. Comparisons were made by gender and ethnic group within undergraduate major field categories. Sample sizes ranged from 1,815 to 1,930 for the 8 new test editions between October 1985 and April 1987. Results suggest that all verbal item types studied exhibited concurrent validity, with only small differences among the item types. All the item types were valid, and they were very highly correlated. Because of this overlapping variance, little concurrent validity is lost by deleting any one item type. Reading comprehension items may, overall, be slightly more valid than the other item types; and the analogy item type may contribute slightly less to the concurrent validity of the verbal measure. However, these small differences do not suggest any specific revisions to the verbal measure of the GRE General Test. An appendix discusses equating the item-type subscores. (Contains 16 tables, 2 figures, and 24 references.) (Author/SLD)