Bacteriostatic activity was measured in 244 specimens of milk collected during 1977 throughout lactation of up to one year from 78 mothers; the activity varied from very good to fair and only seven were inactive. There was a wider range of activity than was found previously in milk from English mothers. Activity usually fell slowly during lactation but some of the Gambian mothers produced milk of very high activity, like that of colostrum into the second week of lactation, and two mothers did so at six and nine months; other mothers produced good-activity milk throughout lactation. The bacteriostatic activity varied little with the season but slight decreases from that expected were found after the high incidence of infant diarrhoea towards the end of the rainy season. The bacteriostatic activity of most of the milk tested could be prevented by iron salts but that of colostrum and some of the milks with high activity could not. Only these highly active colostra and milks were inhibitory in vitro when the inoculum was increased from 10(4) to 10(6) organisms per ml. These and less active milks were able to inhibit the smaller, standard inoculum for longer than 3 h with the addition of bicarbonate and extra iron-binding protein at the concentrations likely to be present in vivo. Both commensal and pathogenic E. coli were inhibited to a similar degree by these milks and there was no evidence of serotype specificity.