The biologic relation between dietary fats and serum cholesterol established in controlled dietary studies usually has not been found in cross-sectional studies of the general population. In vegetarian groups, dietary variables and serum cholesterol have been correlated significantly. To examine the role of technique of dietary assessment versus the dietary pattern of vegetarians, the authors studied the relation of diet with total serum cholesterol in 46 predominantly vegetarian adults in the Boston, Massachusetts, area in 1973-1974. The basis of the dietary information was 10-day diet records. Total serum cholesterol was positively associated with dietary cholesterol (r = 0.53) and saturated fatty acids (r = 0.50) in partial correlation analysis adjusted for age, sex, and triceps skinfold. The use of one-day dietary records lowered these correlation coefficients to about 0.3. Analysis of the components of variation of nutrient intake demonstrated that the vegetarians had a lower within-person variance, a higher between-person variance, or both compared with nonvegetarian study groups. Biologic responsiveness to dietary fat in the vegetarians was similar to that predicted by the Keys equation derived from nonvegetarians. Therefore, multiple-day averaging of dietary records and relatively smaller ratio of within-person to between-person variation in intake favor the detection of cross-sectional associations of diet with serum cholesterol.