The possible protective effect of breast milk against atopic manifestations in infancy, i.e. atopic eczema and food allergy, has been controversial for the last decades. Besides the methodological problems, differences in the composition of human milk could explain these controversies. The aim of this study was to investigate the composition of polyunsaturated fatty acids (PUFA) and secretory immunoglobulin A (S-IgA) levels to food proteins (ovalbumin and β-lactoglobulin) and an inhalant allergen (cat) in milk from mothers of allergic and non-allergic children. Blood samples were obtained at birth and at 3 months from 120 children. Skin prick tests were performed at 6, 12 and 18 months, and the development of atopic diseases was assessed in the children. Breast milk samples were collected from their mothers at birth and monthly during the lactation period. Milk PUFA composition was measured by gas chromatography, and enzyme-linked immunosorbent assay (ELISA) was used to measure total S-IgA, anti-cat S-IgA, anti-ovalbumin S-IgA, and anti-β-lactoglobulin S-IgA. Allergic disease developed in 44/120 children (22/63 children of allergic mothers and 22/57 children of non-allergic mothers). Lower levels of eicosapentaenoic acid, C20:5 n-3 (EPA), docosapentaenoic acid C22:5n-3 (DPA), and docosatetraenoic acid C22:4 n-6 (DHA) (p < 0.05 for all) were found in mature milk from mothers of allergic as compared to milk from mothers of non-allergic children. The total n-6 : total n-3 and the arachidonic acid, C20:4 n-6 (AA) : EPA ratios were significantly lower in transitional and mature milk from mothers of allergic children, as compared to milk from mothers of non-allergic children. The PUFA levels in serum of allergic and non-allergic children were largely similar, except for higher levels of C22:4 n-6 and C22:5 n-6 (p < 0.05 for both) and a higher AA : EPA ratio in serum phospholipids in the former group (p < 0.05). Changes in the levels of milk PUFA were reflected in changes in PUFA serum phospholipids, particularly for the n-6 PUFA. The AA : EPA ratio in maternal milk was related, however, to the AA : EPA only in serum from non-allergic children, while this was not the case in allergic children. The levels of total S-IgA, anti-cat S-IgA, anti-ovalbumin S-IgA, and anti-β-lactoglobulin S-IgA in milk from mothers of allergic, as compared to non-allergic, children were similar through the first 3 months of lactation. Low levels of n-3 PUFA in human milk, and particularly a high AA : EPA ratio in maternal milk and serum phospholipids in the infants, were related to the development of symptoms of allergic disease at 18 months of age. The milk PUFA composition influenced the composition of PUFA in serum phospholipids of the children. We also showed that the lower levels of colostral anti-ovalbumin S-IgA and lower total S-IgA in mature milk from atopic mothers did not influence the development of allergic disease in the children up to 18 months of age. The findings indicate that low α-linolenic acid, C18:3 n-3 (LNA) and n-3 long-chain polyunsaturated fatty acids (LCP) 20–22 carbon chains, but not the levels of S-IgA antibodies to allergens, are related to the development of atopy in children.