The accuracy of 3 measures of future income was investigated for a sample of families filing Parents' Confidential Statements with the College Scholarship Service (CSS). Measures were: parents' reports of current (1968) income; parents' estimates of 1969 income; and predictions of 1969 income made by application of least squares regression parameters to parents' estimate of income. Two types of accuracy were considered: that of prediction assessed by the correlation between current or estimated income and the actual 1969 income. Analysis indicated that the best single predictor was estimated income, followed closely by current income. The accuracy of each income measure was also investigated by assessing median discrepancies between parents' contribution associated with each income measure and the contribution associated with actual 1969 income. Current and estimated income result with about equal frequency in substantial undercalculations of parents' contribution. But use of current rather than estimated income would result in substantial overcalculations for approximately 31,000 CSS families. Needs analysis procedures relying on current or estimated income, therefore, place low income families at a disadvantage by overstating their future year income. (JS)