A random group of 49 items was drawn from nine commerically available reading comprehension tests. Each test was classified independently by two judges as either a measure of the ability to find answers to questions answered explicitly or in paraphrase in the passages, a measure of the ability to draw inferences or deductions, or a measure of some "other" skill. Both judges classified a majority of the items as measures of the ability to draw inferences or deductions, and there was a reasonable amount of agreement between the judges in this classification process. The judges also indicated the extent to which they thought seven types of faults were present in each item. One judge found a total of 122 faults in the 49 items; the other judge found 31. The judges were most often in agreement in judging items to be measures of general knowledge rather than measures of the ability to comprehend specific passages. (Author)