To describe the current practice of hemodialysis in acute renal failure (ARF) and to estimate the impact of hemodialysis modality on patient outcome. Prospective multicenter observational study conducted from March 1996 to May 1997. The 28 multidisciplinary ICUs in the Rhône-Alpes region in France. The 587 patients who required hemodialysis. Patients were followed until hospital discharge. Among the 587 patients 354 received continuous (CRRT) and 233 intermittent (IRRT) renal replacement therapy as first choice. CRRT patients had a higher number of organ dysfunctions on admission and at the time of ARF and higher SAPS II at time of ARF. Mortality was 79% in the CRRT group and 59% in the IRRT group. Logistic regression analysis showed decreased patient survival to be associated with SAPS II on admission, oliguria, admission from hospital or emergency room, number of days between admission and ARF, cardiac dysfunction at time of ARF, and ischemic ARF. No underlying disease or nonfatal disease, and absence of hepatic dysfunction were associated with an increase in patient survival. The type of renal replacement therapy was not significantly associated with outcome. Renal replacement therapy mode was not found to have any prognostic value. Randomized controlled trials should be undertaken to assess this important question.