The primary aim of the high-dose 5-fluorouracil (5-FU) and leucovorin (LV; HDFL48) phase I study was to determine the maximum tolerated dose and dose-limiting toxicity of 5-FU and LV with modified tri-monthly 48-h continuous infusion of high-dose 5-FU/LV in patients with metastatic colorectal cancer. The study also determined the pharmacokinetic parameters of 5-FU, especially steady-state plasma and bone marrow (BM) concentrations. Eligibility included serum triglyceride of more than or equal to 70 mg/dl, adequate BM function, and the major typical trial criteria. Sixteen patients who were enrolled received tri-monthly 5-FU at 2500 mg/m²/48 h/week (with 500 then 250 mg/m²/48 h/week escalation by a conventional 3-3 schema up to 3750 mg/m²/48 h/week) and LV at 300 mg/m²/48 h on days 1, 8, and 15 during a regular 28-day cycle. The maximum tolerated dose of 5-FU was 3750 mg/m²/48 h/week with this tri-monthly schedule. Dose-limiting toxicities were grade III neutropenia with more than 1-week delay of the next cycle, grade III mucositis, diarrhea, and hand-foot syndrome for more than 3 days. The regimen maintains significant concentration differences between steady-state plasma and BM concentrations (8.06 ± 6.39 vs. 2.89 ± 1.01 μmol/l, P=0.021) as measured by high-performance liquid chromatography. Toxicities were of minor grade and were tolerable, with minimal myelotoxicity significantly associated with low steady-state BM concentration. None had 5-FU-related hyperammonemic encephalopathy. Three patients had an objective partial response (18.8%, 95% confidence interval: 4-46%) and two of the 14 patients, who had failed HDFL24 (2600 mg/m²/24 h/week), had a partial response to HDFL48 (14.3% partial response, 95% confidence interval 2-43%). Median progression-free survival and overall survival were 4.1 months (range: 1.8-12.5) and 10.5 months (range: 2.7-32.1), respectively. The efficacy and low myelotoxicity of HDFL48 were attributed to the sustained adequate steady-state plasma concentration and an average 2.63-fold concentration gradient between plasma and BM compartments at steady state. The recommended 5-FU dose for use in future trials was 3500 mg/m²/48 h/week, with a fixed dose of LV at 300 mg/m²/48 h/week.