This paper acknowledges that research on school dropouts presents problems in the definitions and computation of dropout indices. It considers a variety of methods for computing dropout rates, using dropout rates which are real and which were available at the end of the 1986-1987 school year. Fall enrollment, average daily membership, and total population are presented as three ways of measuring enrollment during the 1986-1987 school year. Data from student withdrawal records is categorized under the headings of no-shows, voluntary dropouts, walkaways, and involuntary withdrawals. Five methods are discussed for computing dropout statistics in the area of attendance and non-attendance, and the differences in the dropout rates obtained by each of these methods are described. Two methods are described for computing statistics on dropouts and no-shows, one longitudinal method is presented, and four cohort-longitudinal methods are discussed. Results of different methods are presented in data tables, with table 7 summarizing the dropout rates computed by the 12 methods. Table 7 illustrates that the rates range from a low of 7.1, computed by comparing the change in enrollment from one grade level to the next for 1 year, to a high of 64.7, computed by tracking a cohort over 6 years, through to graduation. It is noted that dropout rates computed over secondary schools as opposed to high schools are generally lower, and that cross-sectional rates are always lower than longitudinal rates. (NB)