The rapid dwindling of classes in the upper grades of U.S. grammar and high schools has been often noted, and many suggestions as to the improvement of the system of education have been emphasized by reference to this tendency. It is clear that after all that has been done, the attendance of pupils, particularly in the elementary schools, is still in an unsatisfactory condition, and this condition becomes more unsatisfactory in the later years of the school course. In order, however, that any safe conclusions can be based upon these facts of current observation, it is necessary that the facts should be more exactly determined and more carefully analyzed. Some important studies have been made in this field within the past ten to fifteen years that have tended to show the danger of unqualified and sweeping statements with reference to the withdrawal of pupils from school, and to show also the difficulty of ascertaining the exact condition of school attendance with the statistical methods and materials available. This document carries this inquiry much further than previous studies. Proceeding by well-defined methods of modern statistical investigation, it shows clearly the necessary limits of the exact knowledge of the subject, but points out also the conclusions which may be drawn from such records as are obtainable. This document is presented in five sections, as follows: (1) Elimination by grades (the amount of elimination, the variability among cities, and the course of elimination); (2) Elimination by ages (the amount of elimination, the variability among cities, and the course of elimination); (3) Special reports (elimination with respect to sex, elimination in schools for colored children, and elimination in colleges); (4) Data on elimination by grades (the original school grade populations, the reliability of the data as representative of the general tendencies of the cities, the process of estimating actual elimination from the facts of grade populations, and additional data); and (5) Elimination by age (the original data of age populations, the reliability of age data from a few years as representative of the general tendencies of cities, the process of estimating actual elimination from the facts of school-age populations, and the variability among cities with respect to elimination by age. An index is also provided. (Contains 19 tables, 9 figures, and 11 footnotes.) [Best copy available has been provided.]