Little is known about the interrelationships between maternal and infant erythrocyte-DHA, milk-DHA and maternal adipose tissue (AT)-DHA contents. We studied these relationships in four tribes in Tanzania (Maasai, Pare, Sengerema and Ukerewe) differing in their lifetime intakes of fish. Cross-sectional samples were collected at delivery and after 3 d and 3 months of exclusive breast-feeding. We found that intra-uterine biomagnification is a sign of low maternal DHA status, that genuine biomagnification occurs during lactation, that lactating mothers with low DHA status cannot augment their infants' DHA status, and that lactating mothers lose DHA independent of their DHA status. A maternal erythrocyte-DHA content of 8 wt% was found to correspond with a mature milk-DHA content of 1·0 wt% and with subcutaneous and abdominal (omentum) AT-DHA contents of about 0·39 and 0·52 wt%, respectively. Consequently, 1 wt% DHA might be a target for Western human milk and infant formula that has milk arachidonic acid, EPA and linoleic acid contents of 0·55, 0·22 and 9·32 wt%, respectively. With increasing DHA status, the erythrocyte-DHA content reaches a plateau of about 9 wt%, and it plateaus more readily than milk-DHA and AT-DHA contents. Compared with the average Tanzanian-Ukerewe woman, the average US woman has four times lower AT-DHA content (0·4 v. 0·1 wt%) and five times lower mature milk-DHA output (301 v. 60 mg/d), which contrasts with her estimated 1·8–2·6 times lower mobilisable AT-DHA content (19 v. 35–50 g).