To evaluate the validity of the cross-check dietary history method for the assessment of the habitual intake of polyunsaturated fatty acids, the relationship was investigated between the estimated intake of these fatty acids and their relative amounts in the various serum lipid fractions. Food intake data were obtained from a group of volunteers participating in a long-term epidemiological study in the Dutch town of Zutphen. The fatty acid composition of total lipids, phospholipids, triglycerides and cholesterol esters was determined in fasting serum samples from these volunteers. Although correlations between the calculated dietary intake of timnodonic acid (20:5(n-3), TA) and cervonic acid (22:6(n-3), CA), on the one hand, and the relative amounts of these fatty acids in the various serum lipid fractions, on the other, appeared to be statistically significant, no more than about 20 per cent of the variance in the serum values could be explained on the basis of dietary intake. For linoleic acid (18:2(n-6), LA) this was 55 per cent. It is concluded that the cross-check dietary history method is an acceptable technique for estimating the habitual intake of linoleic acid of a given individual. For the fish-related fatty acids, however, this method is less reliable. This discrepancy most probably results from the large differences in TA and CA contents in the same foods. In addition, a negative association with the calculated LA intake on the proportional amount of TA for most of the serum lipid fractions was demonstrated. This indicates an interference of dietary LA on the metabolism of TA.