For several decades, school districts have attempted to anticipate and plan for enrollment fluctuations within 5- and 10-year periods. From a variety of humanistic, educational, and budgeting perspectives, total district and school-by-school forecasts and forecasting models must be as accurate as possible. This paper reviews nine historical forecasting techniques, summarizes their major weaknesses, and identifies three modern methods applicable to school districts: the cohort-survival, percentage-survival, and graphic technique methods. Then the enrollment forecasting procedures at San Diego Unified School District (California) are discussed, including an unacceptable error rate for three forecasting years (1978-79, 1979-80, and 1980-81). Unpopular school closures and state-mandated school integration sparked efforts to combine forecasting techniques and information to produce more accurate forecasts. The present models are more precise. The total district forecast for 1982 had an error rate of 34; for 1983, the error rate was 1.9. Increased accuracy is also assured by the adjustment steps instituted for school-by-school enrollments. Planning procedures have been refined. The paper notes four model development considerations and advises districts to supplement formal forecasting methods with informal ones. (MLH)