Many factors have been proposed as predictors of poor renal prognosis in children with hemolytic uremic syndrome (HUS), but their role is still controversial. Our aim was to detect the most reliable early predictors of poor renal prognosis to promptly identify children at major risk of bad outcome who could eventually benefit from early specific treatments, such as plasmapheresis. Prognostic factors identifiable at onset of HUS were evaluated by survival analysis and a proportional hazard model. These included age at onset, prodromal diarrhea (D), leukocyte count, central nervous system (CNS) involvement, and evidence of Shiga toxin-producing Escherichia coli (STEC) infection. Three hundred and eighty-seven HUS cases were reported; 276 were investigated for STEC infection and 189 (68%) proved positive. Age at onset, leukocyte count, and CNS involvement were not associated with the time to recovery. Absence of prodromal D and lack of evidence of STEC infection were independently associated with a poor renal prognosis; only 34% of patients D(-)STEC(- )recovered normal renal function compared with 65%-76% of D(+)STEC(+), D(+)STEC(-) and D(-)STEC(+ )patients. In conclusion, absence of both D and evidence of STEC infection are needed to identify patients with HUS and worst prognosis, while D(-) but STEC(+) patients have a significantly better prognosis.