We assessed the validity of a self-administered semiquantitative food frequency questionnaire by comparing carotene and vitamin E intake scores derived from form with plasma carotenoid and alpha-tocopherol levels among a group of 59 men and women. The simple correlation between carotene intake and plasma carotenoid was 0.29 (p = 0.02) and the partial correlation was 0.35 (p = 0.005) after adjustment for age, sex, total caloric intake, and plasma cholesterol and triglycerides. The simple correlation between vitamin E intake and plasma alpha-tocopherol was 0.12 (p = 0.19). However, adjustment for total caloric intake and plasma lipids each substantially increased this association so that the partial correlation adjusting simultaneously for caloric intake, plasma lipids, age, and sex was 0.34 (p = 0.006). As expected on the basis of previous randomized trials, the correlation between preformed vitamin A intake and plasma retinol was weak and not statistically significant. These data support the utility of self-administered food frequency questionnaires for use in epidemiological studies. In addition, they illustrate the importance of adjusting blood alpha-tocopherol levels for lipid concentrations when the former are used as surrogates for vitamin E intake.