The risk of hemolytic uremic syndrome (HUS) following E. coli O157 gastroenteritis has been estimated to range widely (from 3-26%), and has been reported as a summary estimate for all ages. The objective of this study was to better estimate the age-specific risk of HUS and hemolytic anemia (HA) following O157 infection among a representative cohort of children from the Province of Alberta, and to compare this to children presenting to tertiary centers in the rest of Canada. Children with HUS or E. coli O157 gastroenteritis were eligible if they were less than 15 years old and stool samples had been submitted to one of 18 participating hospital labs or to one of the two Provincial Health Laboratories (PHL) in Alberta. Samples from the PHL allowed identification of an almost complete cohort of non-referred Alberta children with O157 gastroenteritis. Hemoglobin, blood smear, urinalysis, and serum creatinine were obtained 8-10 days after the onset of diarrhea; subjects were followed for 1 month. Compliance with blood specimens was 92% in Alberta and 90% in the rest of Canada. O157 infection was considered present in those with HUS if the stool culture grew O157 or if the acute or convalescent anti-O157 lipopolysaccharide antibody titer was ≥ 1:500. From June, 1991 to March, 1994 HUS was diagnosed in 205 Canadian children. Of these, 77% (26 in Alberta and 131 in other provinces) had evidence of O157 infection. In addition 586 children had O157 gastroenteritis but no HUS; 4 of these in Alberta and 14 elsewhere had HA. The risk of HUS following O157 infection in Alberta was 26/322, or 8.1% (95% CI, 5.3-11.6); if those with HA are included, the risk increases to 9.3%. In the rest of Canada, the risk of HUS following O157 infection was 31.1%, reflecting referral bias. In Alberta children, the highest age-specific risk of HUS/HA was 13% in those younger than 5 years (95% CI, 7.6-18.5%). These data help guide clinical care and provide a basis for sample size estimates for prevention trials. More than 90% of HUS patients across Canada had been seen by a physician in the two weeks before diagnosis, identifying the potential for intervention at an early phase of the illness.