Since 1915, the movement to test reading ability has grown in importance and pervasiveness. Testing is relied on to offer quantitative information, to determine the usefulness of instructional approaches, and to help ensure accountability. The demand for tests arose during the period from 1990 to 1992 when school became compulsory and educators faced new challenges with the range of abilities and diversity of cultures that individuals brought to the public school system. In 1915, W. S. Gray published the first standardized reading test. Later, other reading tests were published to measure comprehension. During the period from 1930 to 1965, schools relied heavily on standardized tests to diagnose disability, predict success, place students in special programs, and demonstrate instructional effectiveness. From 1966 to 1991, federal legislation and judicial decisions requiring public schools to provide free educational services contributed to the further pervasiveness of testing. Although reading tests have not changed significantly in the past 70 years, there have been advances in statistical sophistication relative to reliability and validity. But there is a growing dichotomy between recent advances in understanding the reading process and the tests used to measure reading achievement. The following five recommendations can help to improve large-scale reading assessment: (1) tests should be administered with a specific purpose; (2) the amount of testing should be reduced; (3) knowledge about tests should be improved and increased; (4) standardized tests should be carefully chosen; and (5) assessment information should be used for instructional purposes at a personal level. (One table of data is included and 32 references are attached.) (PRA)