Some studies on the role and behaviour of organic matter during coagulation of raw waters or solutions have been reported in the literature. As far as coagulation of dissolved humic substances is concerned, some general conclusions could be made from the studies. The best removal of these substances was obtained at pH 4–5 with ferric iron and between pH 5 and 6 with alum (Hall and Packham, 1965; Van Breemen et al., 1979; Jekel, 1985; Lefebvre and Legube, 1990). Fulvic acids were found to be less easy to remove than humic acids (Babcock and Singer, 1979; Jekel, 1985). A relationship between initial concentration of humics and optimal coagulant dosage was often shown (Narkis and Rebhun, 1977; Edzwald et al., 1977, 1979; Kim et al., 1989; Lefebvre and Legube, 1990). The aim of our study was to optimize the removal of organic matter by iron(III) coagulation. A first set of experiments on aquatic humic substances allowed some results to be obtained (Lefebvre and Legube, 1990) such as an optimum of pH between 4.5 and 5.5 and the presence of the stoichiometric relationship of 2 mg Fe per mg of DOC, to obtain the best removal at pH 5.5. The objective of this paper was to try to define the role of the acidic function of organic matter on its removal by iron(III). Some simple aromatics and aliphatic acids were studied in weakly mineralized solution. Experiments were carried out with laboratory apparatus including coagulation-flocculation sedimentation and flocculation treatment steps (Lefebvre and Legube, 1990). The removal of organic compounds was followed by TOC and u.v.-absorbance analysis. Residual iron was determined by atomic adsorption spectrometry. The optimal pH of coagulation was found to be about 5.8 for the salicylic acid and 5.5 for gallic acid. Figure 1 shows data obtained with gallic acid. Hence, the following studies were performed at pH 5.5. Furthermore, choice of this pH allowed comparison with results obtained from coagulation of fulvic acids in similar conditions (Lefebvre and Legube, 1990). Table 2 collects average values of results obtained with 17 aromatic compounds (2.5–15 mg l −1 TOC) coagulated at pH = 5.5 and 20°C, with 15, 30 or 40 mg l −1 Fe. Examination of Table 2 allows us to define four classes of aromatic compounds regarding their behaviour during iron coagulation. Aromatics bringing only one acidic group (COOH and OH) or more (two or three) non-adjacent acidic groups, were not removed under our experimental conditions (benzoic acid, p-hydroxybenzoic acid, m-hydroxybenzoic acid, 3-methoxybenzoic acid, o-aminobenzoic acid, phenol, resorcinol, phoroglucinol, o-cresol). Aromatics bringing at least two adjacent acidic groups were more or less removed. The type of acid group seemed to play a significant role on the removal yield. Two different groups in the ortho position led to a small removal not depending on the coagulant/influent TOC ratio, as shown by Fig. 2 for salicylic acid. Two identical groups in the ortho position induced an increasing removal when the coagulant/TOC i ratio increased. Figure 3 illustrates this in the case of phthalic acid. Finally, two adjacent acidic groups on an aromatic compound led to a similar behaviour for humic substances (Lefebvre and Legube, 1990), that is the presence of optimum removal when the coagulant/TOC i ratio varied (gallic acid, pyromellitic acid and pyrogallol). Figure 4 shows an example of gallic acid. Figure 5 summarizes general shapes of curves in terms of remaining compound (TOC effluent/TOC influent) versus coagulant dose (mole/mole). Table 3 gives the values of the results obtained for some aliphatic diacids, coagulated under similar conditions (pH = 5.5, 20°C, TOC i = 2–16 mg l −1, Fe = 30 mg l −1). Figure 6 presents the synthesis. In the case of the saturated aliphatic, removal depended on the number ( n) of CH 2 groups between the two acidic functions. The smaller n was, the better the removal of diacid. Discussion about the possible mechanisms concludes on two main pathways. The first is an exchange of ligand to the surface of amorphous FeOOH. The second is a complexation or ligand exchange between soluble monomers, dimers or oligomers of iron hydroxide and organic matter to lead (directly or after hydrolysis) to a precipitate.