The accuracy of Bureau of Labor Statistics (BLS) projections of occupational employment growth was investigated for the period 1972-1989. BLS was found to have consistently underpredicted the growth of skilled occupations and overpredicted the growth of occupations requiring low or moderate skills. A regression-based forecast was indicated as doing a better job of forecasting future occupational staffing ratios than the judgmental approach used by BLS. A regression analysis of the substantial changes in occupational shares that occurred from 1972 to 1989 predicted substantially greater upskilling than BLS. BLS's projections of the supply/demand balance for college graduates were also evaluated and found to have been consistently wrong. An examination of this balance was done by comparing past and projected percentage rates of change in employment in high skill jobs to actual and projected percentage rates of change in the stock of well-educated workers. Projections indicated that the labor market for college graduates was tight and would get even tighter. Findings implied high social returns to a college education and the need for a public policy response to ameliorate the shortage. (Appendixes include 14 tables and 42 endnotes.) (YLB)