Acute renal failure (ARF) in critically ill patients is associated with high mortality. Optimal method and dose of continuous renal replacement therapy could improve survival in these patients. We studied the hypothesis that an increase in dialysis dose obtained by continuous veno-venous hemodiafiltration (CVVHDF) is associated with a better survival than continuous veno-venous hemofiltration (CVVH) among critically ill patients with ARF. In a prospective randomized trial, these two methods were compared in patients undergoing renal replacement therapy in two intensive care units (ICUs). The patients had either CVVH (1-2.5 l/h replacement fluid) or continuous CVVHDF (1-2.5 l/h replacement fluid+1-1.5 l/h dialysate) according to their body weight. 28- and 90-day mortalities, renal recovery, and duration of ICU stay were the main outcome measures. Two hundred and six patients were randomized from October 2000 to December 2003. Twenty-eight-day survivals (%) were, respectively, 39 and 59 (P=0.03) in the CVVH and CVVHDF groups. Three months survivals (%) were, respectively, 34 and 59 (P=0.0005) in the CVVH and CVVHDF groups. Apache II score, age, baseline blood urea nitrogen, and hemodiafiltration (hazard ratio 0.59, 95% confidence interval 0.40-0.87; P=0.008) were independent predictors of survival at 90 days. Renal recovery rate among survivors (71 versus 78% in the CVVH and CVVHDF groups respectively, P=0.62) was not affected by the type of renal replacement therapy. These results suggest that increasing the dialysis dose especially for low molecular weight solutes confers a better survival in severely ill patients with ARF.