Dropout rates are used by some experts as educational indicators of quality and to gauge the holding power of special education programs. Despite the difficulty in comparing data from different agencies, the evidence suggests higher dropout rates for children with mild disabilities, particularly learning and emotional handicaps. The authors suggest that teachers and administrators should be familiar with characteristics of students and of schools that place children at risk of dropping out. Efforts to reduce the number of students require the establishment of reliable and valid estimates of the magnitude of the problem. Dropout prevention programs must consider the reasons students leave, which are not necessarily the same as those factors that predict early school leaving. 5 tables and 43 references