This study evaluated the reliability and comparative validity of the dietary history survey method developed for CARDIA, a longitudinal investigation of life-styles and the evolution of cardiovascular risk factors in young adults. The method was tested in a sample of 30 white men, 33 white women, 33 black men, and 32 black women, aged 18 to 35 years, in four regions reflecting the race, sex, age, and geographical distribution of CARDIA participants. For 64 of the participants, two dietary history interviews, 1 month apart, were conducted by trained nutritionists. These data were used to examine the reliability of the method. For all participants, seven telephone- assessed 24-hour dietary recalls were randomly scheduled during a 28-day period and were followed by a dietary history interview. These data were used to examine the comparative validity of the method. Calories, total fat, saturated fat, polyunsaturated fat, monounsaturated fat, dietary cholesterol, protein, carbohydrate, alcohol, potassium, calcium, and vitamin A were selected by the CARDIA Nutrition Working Group as the nutrients for comparison. Mean nutrient values from the first history tended to be higher than those obtained from the last history. However, for a majority of the nutrients, the differences were significant for blacks but not for whites. The correlations for the log-transformed nutrient values and calorie-adjusted nutrient values from the two histories were generally in the range of 0.50 to 0.80 for whites. For blacks, the correlations were lower, with a majority in the range of 0.30 to 0.70. The average nutrient values estimated from the histories were higher than those estimated from the average of the seven 24-hour recalls. The differences in blacks were larger than in whites. For both the log-transformed nutrient values and the calorie-adjusted nutrient values, the correlations between the two methods were generally larger than 0.50 for whites. However, for blacks, the correlations were lower, and for several macronutrients, the correlations were close to zero. These results suggest that the CARDIA dietary history is a reasonably reliable and valid dietary survey method for obtaining information about habitual intakes in whites. In blacks, the results are less consistent. The black-white differences remained when the analyses were stratified by education. These results raise a question about differences in reporting between blacks and whites.