Salmon oil (16·5 kg /t), a source of long-chain polyunsaturated n-3 fatty acids, was included in diets offered to multiparous sows during pregnancy and lactation to measure responses in pre-weaning mortality and performance of piglets in two studies. The first study, carried out under commercial conditions, included 196 sows which were offered salmon oil and control diets from immediately post service until weaning. The same diets were also offered to 10 sows per treatment from day 58 of pregnancy in a controlled nutritional study which measured the effects of salmon oil on piglet tissue fatty acid composition. Offering salmon oil to the sow significantly increased gestation length and decreased individual piglet birth weight but had no effect on litter size at birth. Overall, salmon oil reduced pre-weaning mortality from 11·7% to 10·2% mainly by reducing the incidence of deaths from crushing by the sow. More detailed analysis of mortality using a general linear mixed model and 2294 piglet records, demonstrated that the incidence of pre-weaning mortality was significantly decreased with increasing individual piglet birth weight and by inclusion of salmon oil in the diet; the incidence of mortality increased with average piglet birth weight in a litter. Salmon oil inclusion had no effect on weight of litter weaned, sow lactation food intake or subsequent reproductive performance. In both studies, dietary salmon oil increased the proportions of long-chain n-3 polyunsaturated fatty acids in colostrum to a similar extent. In the nutritional study, inclusion of salmon oil reduced the proportions of 20: 4 n-6 in piglet liver and brain at birth and increased the proportions of long-chain n-3 polyunsaturated fatty acids. Therefore, despite reducing piglet birth weight, offering sows salmon oil reduced pre-weaning mortality of piglets. The nutritional study showed that the amount and type of marine oil used may not have been optimal.