For an evaluation of fat-soluble vitamin concentrations in adipose tissue as biomarkers of intake, estimates of usual intake of beta-carotene, total vitamin A, and vitamin E (assessed by food frequency questionnaire) were compared with plasma and adipose tissue concentrations of beta-carotene, retinol, and alpha-tocopherol, respectively. Data were collected in 1992 in the Netherlands for 85 healthy, nonsmoking volunteers aged 50-70 years (38 males and 47 females). For alpha-tocopherol, a significant age- and sex-adjusted partial correlation (r = 0.24, p < 0.05) was observed between adipose tissue levels and intake. For beta-carotene, the partial r was 0.20. Adipose tissue retinol did not reflect intake (partial r = 0.08). Correlations of adipose tissue vitamin levels with plasma vitamin levels were higher overall (r = 0.34 for alpha-tocopherol, r = 0.56 for beta-carotene, and r = 0.17 for retinol) than correlations with intake. Plasma concentrations of alpha-tocopherol, beta-carotene, and retinol were not associated with dietary intake (partial r's were 0.05, 0.17, and -0.12, respectively). Pearson correlations of repeated measurements in adipose tissue (after 4 months) were 0.24 for retinol, 0.50 for beta-carotene, and 0.78 for alpha-tocopherol. Adipose tissue beta-carotene was shown to increase sixfold after 6 months' supplementation with 30 mg of beta-carotene daily. It is concluded that adipose tissue vitamin concentrations are an acceptable alternative to plasma levels as relatively stable indicators of dietary intake. However, both plasma and adipose tissue levels are more useful as markers of internal dose, taking into account variations in absorption and metabolism, than of dietary intake.