The important of the educational diagnosis in instructing patients in a fat modified diet was examined in a double blind study. The educational diagnosis depended upon the assessment of three independent instructional parameters: the amount of structure, the teaching mode, and the language code preferred by the patient. Group I (67 patients) received traditional personalized counseling for 3 months. Group IIa (40 patients), whose instruction was matched to their preferred education attributes, was randomized against group IIb (36 patients), whose instruction was mismatched. Group IIIa (33 patients), receving matched instruction was randomized against group IIIb (35 patients), receiving high structure in the symbolic mode of instruction. The greatest reduction in serum cholesterol was seen in group I. At 3 months, no significant difference in serum cholesterol reduction could be shown between groups IIa and IIb, or IIIa and IIIb. However, the mean cholesterol reduction became significantly greater (P less than 0.05) in IIa than IIb at the 6 month visit. Seven day food diaries showed no significant differences in dietary intake and 1 and 3 months between matched and unmatched groups. The data did suggest that patients preferring high structure in the symbolic mode had a greater serum cholesterol reduction that other diagnostic groups. Analysis of variance showed an interaction between the preferred and instructional level of structure as related to the change in total fat, and saturated fat of the diet after 3 months.