Many different analytical procedures for fatty acid analysis of infant formulae and human milk are described. The objective was to study possible pitfalls in the use of different acid-catalyzed procedures compared to a base-catalyzed procedure based on sodium-methoxide in methanol. The influence of the different methods on the relative fatty acid composition (wt% of total fatty acids) and the total fatty acid recovery rate (expressed as % of total lipids) was studied in two experimental LCP-containing formulae and a human milk sample. MeOH/HCl-procedures were found to result in an incomplete transesterification of triglycerides, if an additional nonpolar solvent like toluene or hexane is not added and a water-free preparation is not guaranteed. In infant formulae the low transesterification of triglycerides (up to only 37%) could result in an 100%-overestimation of the relative amount of LCP, if these fatty acids primarily derive from phospholipids. This is the case in infant formulae containing egg lipids as raw materials. In formula containing fish oils and in human milk the efficacy of esterification results in incorrect absolute amounts of fatty acids, but has no remarkable effect on the relative fatty acid distribution. This is due to the fact that in these samples LCP are primarily bound to triglycerides. Furthermore, in formulae based on butterfat the derivatization procedure should be designed in such a way that losses of short-chain fatty acids due to evaporation steps can be avoided. The procedure based on sodium methoxide was found to result in a satisfactory (about 90%) conversion of formula lipids and a reliable content of all individual fatty acids. Due to a possibly high amount of free fatty acids in human milk, which are not methylated by sodium-methoxide, caution is expressed about the use of this reagent for fatty acid analysis of mothers milk. It is concluded that accurate fatty acid analysis of infant formulae and human milk requires a careful and quantitative derivatization of both polar and nonpolar lipid classes. Sodium methoxide seems to be a reliable and time-saving method for routine fatty acid analysis of infant formulae, which should be validated by interlaboratory comparison. Anhydrous procedures based on methanolic hydrogen chloride including an additional nonpolar solvent are also suitable for infant formulae but seem to be preferable for human milk samples.