Three methods of estimating vitamin A and C intakes from food frequency data obtained in June 1981 were compared with estimates obtained from dietary histories taken in September 1982, in 50 elderly residents of a Southern California retirement community. The first method of estimating vitamin A and C intakes (Af and Cf) was to add the products of the frequency of use of foods and the vitamin content of an average serving. The second method of estimating the intakes of these vitamins (Ai and Ci) was to develop an index based on the sum of frequencies of consumption of foods rich in these vitamins. This was then converted to absolute amounts using a regression equation. The third method (Ar and Cr) was to use stepwise multiple regression with the estimated intakes from the history (Ah and Ch) as dependent variables and to develop an equation with a small number of foods as the independent variables. When intakes from food alone were examined, the Spearman correlations between Ah and Af, Ai, and Ar, respectively, were 0.03, 0.16, and 0.43; the correlations between Ch and Cf, Ci, and Cr were 0.29, 0.36, and 0.38. When supplemental vitamin intakes were included, the correlations were between 0.35 and 0.44 for vitamin A and between 0.62 and 0.64 for vitamin C. The results demonstrated the importance of including the contribution of supplemental vitamins in estimating total vitamin A and C intake and suggested that the index and regression approaches may be superior to Af and Cf for estimating dietary vitamin intakes from food frequency questionnaires.