This study sought to investigate whether a "natural diet" (mimicking the fatty acid composition of freshwater aquatic insects eaten by salmon parr) during the freshwater (FW) life stage of pre-smolt Atlantic salmon (Salmo salar L.) affected red blood cells and gill fatty acid composition as well as eicosanoid metabolism in gill during smolting at different temperatures. Before being transferred to seawater (SW), salmon parr were fed with a modified (MO) diet containing vegetable oils (rapeseed, palm, and linseed oils) supplemented with eicosapentaenoic acid (EPA) and arachidonic acid (ARA) to completely replace the fish oil (FO). Fatty acid composition in red blood cells and gill tissues was determined before SW transfer and six weeks after. Additionally, the expression of genes associated with eicosanoid metabolism and Na/K-ATPase (NKA) activity in salmon gill was examined at different temperatures before SW transfer and 24 h after. The results showed the changes in fatty acid composition, including sum monounsaturated fatty acids (MUFAs), docosahexaenoic acid (DHA), ARA, EPA, and sum n-6 polyunsaturated fatty acids (n-6 PUFA) in both red blood cells and gill tissues at the FW stage were consistent with the fatty acid profiles of the supplied MO and FO fish diets; however sum EPA and DHA composition exhibited opposite trends to those of the FO diet. The proportion of ARA, EPA, and n-6 PUFA increased, whereas sum MUFAs and DHA decreased in the red blood cells and gill tissues of MO-fed fish compared to those fed with the FO diet at FW stage. Additionally, 5-lipoxygenase-activating protein (Flap) expression was downregulated in MO-fed fish prior to SW transfer. During the process of SW transfer at different temperatures, the MO diet remarkably suppressed NKAα1a expression in MO-fed fish both at 12 and 16 °C. The MO diet also upregulated phospholipase A2 group IV (PLA2g4) expression in gills at 8, 12, and 16 °C, but suppressed phospholipase A2 group VI (PLA2g6) expression in gills at 12 °C compared to FO-fed fish at 12 °C and MO-fed fish at 8 °C. The MO diet also upregulated Cyclooxygenase 2 (Cox-2) expression at 8 °C compared to FO-fed fish and increased Arachidonate 5-lipoxygenase (5-Lox) expression in MO-fed fish at 16 °C compared to both FO-fed fish at 16 °C and MO-fed fish at 8 °C. Our study also determined that both SW transfer water temperatures and diets during the FW period jointly influenced the mRNA expression of PLA2g4, PLA2g6, and Lpl, whereas 5-Lox was more sensitive to dietary changes. In conclusion, the MO diet affected the fatty acid composition in gill and in red blood cells. When transferred to SW, dietary ARA supplementation could promote the bioavailability for eicosanoid synthesis in gill mainly via PLA2g4 activation, and potentially inhibit the stress and inflammatory response caused by different water temperatures through dietary EPA supplementation.