A dairy herd associated with Escherichia coli O157 infection in humans was studied for the 15 months following the outbreak to examine seasonal, age and management factors affecting faecal excretion of the organism and to determine the mode and frequency of milk contamination with the organism. Between May 1993 and July 1994, 28 visits were made to the farm to collect a total of 3593 rectal swabs from cows, heifers and calves and 329 milk samples. E.coli O157:H7 was isolated from 153 (4·3%) of 3593 bovine rectal swabs. The maximum prevalence at any one visit was 14% in lactating cows, 40% in non-lactating cows, 56% in calves and 68% in heifers. The prevalence in lactating cows, which was significantly lower than in the other groups, peaked during May–July 1993 and again briefly after the cattle were housed during November 1993 and then again during May 1994. Excretion rates of E. coli O157:H7 in lactating cows were highest during the first month after calving, falling during lactation and rising to another peak at 7 months postpartum. Between November 1993 and May 1994 there was no evidence of excretion in any group. Eighty-seven (74%) of the animals which excreted E. coli O157:H7 did so on only one occasion but 23 (32%) of 73 cows and heifers and 7 (16%) of 44 calves which excreted the organism did so on more than one occasion. E. coli O157:H7 was not isolated from milk taken from the bulk tank but it was isolated from individual milk samples (one milk jar and one fore-milk) from two animals previously shown to be faecal excretors of the organism. All isolates of E. coli O157:H7 obtained were of the same phage type, toxin genotype and plasmid profile.