Duplicate groups of Atlantic salmon post-smolts were fed five practical-type diets in which the added oil was 100% fish oil (FO)/0% rapeseed oil (0% RO), 90% FO/10% RO (10% RO), 75% FO/25% RO (25% RO), 50% FO/50% RO (50% RO) or 100% RO for 16 weeks. After sampling, the remaining fish were switched to a commercial grower diet containing FO as the only added lipid for a further 12 weeks. There were no effects of the inclusion of RO on growth or feed conversion. Fatty acid compositions of flesh total lipid showed significant increases in 18:1n-9, 18:2n-6 and 18:3n-3 with increasing inclusion of RO. The percentages of flesh 20:5n-3 and 22:6n-3 were significantly reduced in fish fed 10%, 25% and 50% RO, compared to fish fed 0% RO, and the percentages in fish fed 100% RO were significantly lower than all other treatments. Following transfer to the FO washout diet, percentages of flesh 20:5n-3 were not significantly different among treatments after 4 weeks while percentages of 22:6n-3 were restored after 12 weeks. However, even after 12 weeks, percentages of flesh 18:2n-6 were still significantly higher in fish previously fed 50% and 100% RO compared to the other treatments, although the final values were reduced by 48% and 65%, respectively, following the 12-week washout period. This study suggests that RO is a potential substitute for FO in Atlantic salmon culture, but that percentages of the n-3 highly unsaturated fatty acid (HUFA), 20:5n-3 and 22:6n-3, are significantly reduced by feeding RO above 50% of added oil. However, percentages of these two HUFA can be restored by feeding a diet containing FO for a period of 12 weeks. (C) 2003 Elsevier Science B.V. All rights reserved.