To compare the efficacy, safety, and cost of fixed-dose low-molecular-weight heparin (dalteparin) with adjusted-dose unfractionated heparin as anticoagulant for continuous hemofiltration. Prospective, randomized, controlled clinical trial. University-affiliated adult intensive care unit All patients requiring continuous hemofiltration for acute renal failure or systemic inflammatory response syndrome (SIRS) were eligible. Fifty-seven patients were enrolled. Eleven were excluded, seven because of major protocol violations and four died before hemofiltration. Patients received continuous venovenous hemodialysis with filtration with prefilter replacement at 500 mL/hr and countercurrent dialysate at 1000 mL/hr. Filters were primed with normal saline containing anticoagulant. Dalteparin-treated patients received a commencement bolus of 20 units/kg and a maintenance infusion at 10 units/kg/hr. Heparin-treated patients received a commencement bolus of 2000-5000 units and a maintenance infusion at 10 units/kg/hr, titrated to achieve an activated partial thromboplastin time in the patient of 70-80 secs. The primary outcome measure--time to failure of the hemofilter--was compared using survival analysis. Twenty-two patients (13 with acute renal failure and nine with SIRS; total, 41 filters) were randomized to heparin. Twenty-five patients (16 with acute renal failure and nine with SIRS; total, 41 filters) were randomized to dalteparin. Mean (SE) activated partial thromboplastin time in the heparin group was 79 (4.3) secs. Mean (SE) anti-factor-Xa activity in the six patients given dalteparin who were assayed was 0.49 (0.07). Mean (SE) prehemofiltration platelet count was 225 (35.5) x 10(9) for heparin and 178 (18.1) x 10(9) for dalteparin (p = .24, unpaired Student's t-test). Mean (SE) prehemofiltration hemoglobin was 11.4 (0.61) g/dL for heparin and 10.6 (0.38) g/dL for dalteparin (p = .31, unpaired Student's t-test). There was no significant difference in the time to failure between the two groups (p = .75, log rank test). For dalteparin, Kaplan-Meier (K-M) mean (SE) time to failure of the hemofilter was 46.8 (5.03) hrs. For heparin, K-M mean (SE) time to failure was 51.7 (7.51) hrs. The 95% CI for difference in mean time to failure was -13 to 23 hrs. The power of this study to detect a 50% change in filter life was >90%. Mean (SE) reduction in platelet count during hemofiltration was 63 (25.8) x 10(9) for heparin and 41.8 (26.6) x 10(9) for dalteparin (p = .57, unpaired Student's t-test). Eight patients given dalteparin and four patients given heparin had screening for heparin-induced thrombocytopenia; three of the dalteparin patients and one of the heparin patients were positive (p = 1.0, Fisher's exact test). There were three episodes of trivial bleeding and two episodes of significant bleeding for dalteparin, and there were three episodes of trivial bleeding and four episodes of significant bleeding for heparin (p = .53, chi-square test). The mean (SE) decrease in hemoglobin concentration during hemofiltration was 0.51 (0.54) g/dL for heparin and 0.28 (0.49) g/dL for dalteparin (p = .75, unpaired Student's t-test). The mean (SE) packed-cell transfusion volume during hemofiltration was 309 (128) mL for heparin and 290 (87) mL for dalteparin (p = .90, unpaired Student's t-test). Daily costs, including coagulation assays, of hemofiltration were approximately 10% higher using dalteparin than with heparin. Fixed-dose dalteparin provided identical filter life, comparable safety, but increased total daily cost compared with adjusted-dose heparin. Unfractionated heparin remains our anticoagulant of choice for continuous hemofiltration in intensive care.