Two methods were compared for the determination of free fatty acids (FFA) from acetic to long-chain acids in samples with a large excess of triacylglycerols (TG) (1[ratio ]200, w/w), such as cheese and other dairy products. In method 1, after fat extraction, FFA were separated from TG by aminopropyl-bonded phase chromatography, injecting the fraction containing FFA directly into the gas chromatograph. In method 2, extracted fat was treated with tetramethylammonium hydroxide, the methyl ester derivatives being formed in the injector. Cheese samples and standard mixtures of FFA and TG in different proportions were analysed by both methods. The cheese sample contained 2·4 times more FFA when analysed by method 2 as compared with the result obtained with method 1. The composition of the standard mixtures analysed by method 1 closely reflected that of the original mixture and gave 90–100% recovery of FFA, regardless of their chain length and the ratio of FFA[ratio ]TG (1[ratio ]1 or 1[ratio ]200, w/w). The composition of samples with a FFA[ratio ]TG ratio of 1[ratio ]200 (w/v) was severely distorted (as compared with the original composition of the sample) when analysed by method 2. Varying recoveries of FFA were also obtained, the largest differences being found for the shorter-chain components. We conclude that the FFA fraction should be separated from the TG fraction before derivatization and chromatographic analysis, particularly for samples in which the FFA represent a minor fraction of the TG.