OBJECTIVE To evaluate, quantify and compare the effects of continuous veno-venous hemofiltration (CVVH) with lactate or bicarbonate-buffered replacement fluids on acid-base balance. METHODS Randomized double crossover study. METHODS Intensive Care Unit of Tertiary Medical Center. METHODS Eight patients with severe acute renal failure. METHODS Random allocation to either 2 hours of isovolemic lactate-buffered (treatment A) CVVH or 2 hours of bicarbonate-buffered (treatment B) CVVH with cross over and with same procedure repeated the following day (double cross over). RESULTS Timed collections of arterial blood and ultrafiltrate (UF), measurement of blood and UF gases and lactate concentrations and calculation of buffer-base mass balance. At baseline, both groups of patients had a similar, slight metabolic alkalosis (pH: 7.45 vs. 7.45; BE 3.9 mEq/L for treatment A and 4.0 for treatment B) and a serum bicarbonate of 28.1 mmol/L for treatment A vs. 28.3 mmol/L for treatment B; all NS. This alkalosis was present despite slight hyperlactatemia in both groups (A: 2.4 mmol/L vs. B 2.8 mmol/; NS). Within 60 minutes of treatment, however, treatment A led to a significantly higher lactate concentration (3.9 vs 2.5 mmol/L; p = 0.0011), a significantly lower BE (2.3 vs 4.1 mEq/L; p = 0.0019) and a significantly lower bicarbonate concentration (26.7 vs. 28.3 mmol/L; p = 0.0038) in the presence of an unchanged PaCO2. These differences persisted during the study period. The UF of patients receiving treatment A contained more lactate (10.2 vs 2.9 mmol/L; p < 0.0001) and less bicarbonate (25.6 vs. 30.8 mmol/L; p < 0.0001) than treatment B resulting in a mean buffer-base balance of +20.4 mEq/h compared to -2.6 mEq/h for treatment B; p < 0.0001). CONCLUSIONS CVVH with lactate-buffered replacement fluids induces iatrogenic hyperlactatemia. Such hyperlactatemia is associated with an acidifying effect despite a positive buffer-base balance.