Six cannulated wether sheep weighing 57 (s.d. 4·3) kg were used to investigate the susceptibility of unprotected and protected n-3 polyunsaturated fatty acids from different sources to biohydrogenation in the rumen, their uptake into plasma and effects on ruminal metabolism. The sheep were assigned to one of six dietary treatments formulated to have a similar fatty acid content (60 g/kg DM) and containing: linseed oil (LO), linseed oil absorbed into vermiculite (VLO), formic acid-formaldehyde treated whole linseed (FLS), fish oil (FO), fat encapsulated fish oil (PFO) or a mixture of fish oil and marine algae (1: 1 on an oil basis; AF), in six periods of 28 days duration in a Latin-square design. Biohydrogenation of C20:5 (n-3) and C22:6 (n-3) was high in FO at approximately 870 g/kg, but reduced to 625 and 625 g/kg respectively for PFO, and 769 and 601 g/kg respectively for AF. Ruminal biohydrogenation of C18:3 (n-3) was similar across treatments based on linseed, averaging 860 g/kg, but C18:2 (n-6) was lower (P < 0·05) in animals given VLO or FLS at 792 and 837 g/kg respectively, compared with LO (907 g/kg). Duodenal flow of C18:1 trans in animals given any of the diets containing fish oil averaged 8·4 g/day compared with 2·8 g/day in animals given diets based on linseed (P < 0·001), whilst cis-9, trans-11 conjugated linoleic acid was not significantly different among treatments. Plasma C20:5 (n-3) and C22:6 (n-3) proportions were highest in animals given the AF diet (11·8 and 8·2 g per 100 g of the total fatty acids respectively) and lowest in animals given LO (2·8 and 2·7 g per 100 g of the total fatty acids respectively; P < 0·001). By contrast, plasma C18:3 (n-3) proportions were highest in animals given the LO or VLO diets at approximately 6·9 g per 100 g of the total fatty acids, and lowest in the AF treatment at 0·9 g per 100 g (P < 0·001). Duodenal non-ammonia-N flow was similar among treatments at 21·0 g/day except in animals given FLS which had the highest flow (25·9 g N per day; P < 0·01). Microbial N flow was also similar among treatments whilst microbial efficiency (g N per kg OM truly degraded in the rumen) was higher (P < 0·05) in animals given FLS than LO, FO or AF. By contrast, ruminal fibre digestion was higher (P < 0·05) in animals given LO or FO than those offered VLO, FLS, PFO or AF. In conclusion, compared with linseed oil, absorption of linseed oil into vermiculite improved duodenal flow but not plasma levels of C18:3 (n-3), whilst formic acid-formaldehyde treatment of linseed had little effect on protecting C18:3 (n-3) in the rumen, although duodenal non-ammonia nitrogen flow and microbial efficiency were improved. Compared with fish oil, the provision of marine algae or fat encapsulated fish oil resulted in a lower biohydrogenation of C22:6 (n-3) and C20:5 (n-3), and an increased duodenal flow and plasma concentration and offers the potential to favourably manipulate the n-3 fatty acid composition of sheep meat.