The associations of retinol and beta-carotene plasma concentrations with eight personal variables and the use of seven classes of cardiovascular drugs were studied in over 1,750 patients with nonmelanoma skin cancer enrolled at four American study centers in a cancer prevention clinical trial. Dietary carotene and female sex were positively related to beta-carotene levels, while cigarette smoking and Quetelet index were negatively related. Use of vitamins, beta blockers, or other antihypertensive drugs were also related to beta-carotene levels, but were associated with much smaller changes in these levels. Age and use of other types of cardiovascular drugs were not associated with beta-carotene levels to a statistically significant extent. There was no statistically significant interaction of smoking and dietary carotene in predicting plasma beta-carotene levels. The multiple correlation coefficient between log plasma beta-carotene and the full model was R = 0.50. Retinol levels were positively related to male sex and use of vitamins, diuretics, beta blockers, other cardiovascular drugs, and menopausal estrogens, and negatively related to current cigarette smoking and use of nitrates. The multiple correlation coefficient between plasma retinol and the full model was R = 0.33. These findings confirm the importance of several previously reported predictors of plasma retinol and beta-carotene levels. They also identify several new predictors of these micronutrient levels.