To compare the efficacy and safety of adjusted-dose unfractionated heparin with that of regional citrate anticoagulation in intensive care patients treated by continuous venovenous hemofiltration (CVVH). Prospective, randomized, clinical trial in a 32-bed medical and surgical ICU in a university teaching hospital. ICU patients with acute renal failure requiring continuous renal replacement therapy, without cirrhosis, severe coagulopathy, or known sensitivity to heparin. Before the first CVVH run patients were randomized to receive anticoagulation with heparin or trisodium citrate. Patients eligible for another CVVH run received the other study medication in a cross-over fashion until the fourth circuit. Forty-nine circuits (hemofilters) were analyzed: 23 with heparin and 26 with citrate. The median lifetime of hemofilters was 70 h (interquartile range 44-140) with citrate anticoagulation and 40 h (17-48) with heparin (p=0.0007). One major bleeding occurred during heparin anticoagulation and one metabolic alkalosis (pH=7.60) was noted with citrate after a protocol violation. Transfusion rates (units of red cells per day of CVVH) were, respectively, 0.2 (0.0-0.4) with citrate and 1.0 (0.0-2.0) with heparin (p=0.0008). Regional citrate anticoagulation seems superior to heparin for the filter lifetime and transfusion requirements in ICU patients treated by continuous renal replacement therapy.