This paper develops a revised estimate of the number of children eligible for bilingual services based on the contention that the figure being used by the Department of Education is an overestimate. The current estimate of 3.6 million, which was derived from the Children's English and Services Study (CESS), is believed to be based on excessively high cutoff scores on the CESS test of English proficiency, and to include children who may not be entitled to services because they do not depend on a non-English language or becuase they are not in school. Drawing on legislative and court provisions for bilingual education, the paper explores alternative definitions of the following criteria to delimit the eligible populations: base populations; sufficient limitation in English; and dependence on a non-English language. Using the definitions, the paper generates new estimates by: (1) reanalyzing the CESS data to exclude nonrelevant populations; (2) incorporating a more realistic criterion for limited English proficiency derived from the Sustaining Effects Study (SES); (3) extrapolating estimates to the school universe using Survey of Income and Education (SIE) data; and (4) comparing derived estimates with estimates from other sources. The revised estimates are considered more reliable and more reasonable for Federal policy purposes. (Author/MJL)