Efforts to improve efficacy and minimize toxicity have led to pharmacokinetic monitoring of plasma 5-Fluorouracil (5-FU) levels in colorectal cancer patients undergoing chemotherapy. We observed variation in basal 5-FU levels in 21 patients and significant variation during subsequent dose optimization. Tumor KRAS, BRAF mutations and TS mRNA levels were determined. Regimens included FOLFOX6 + Avastin (N = 8), FOLFOX6 (N = 11), FOLFIRI (N = 1) and FOLFOX4 (N = 1). Mutations identified in tumors included G12V KRAS (N = 2), G12A KRAS (N = 1), and V600E BRAF (N = 3). Six-of-eleven patients with normalized tumor TS mRNA levels < 4.0 had a 5-FU AUC of 20 mg.h/L or greater, and 80% of patients (4 of 5) with TS levels > 4.0 had a plasma 5-FU AUC of less than or equal to 20 mg.h/L. Approximately 2/3 of patients achieved therapeutic 5-FU AUC levels with 0-2 dose adjustments while a sub-group of ~1/3 of patients slowly achieved therapeutic levels (> 3-4 dose increases leading to supra-therapeutic 5-FU and subsequent reductions to lesser than original doses). Liver metastases and tumor TS levels did not fully account for variable 5-FU AUC optimization patterns. The 5-FU level during continuous infusion was half-therapeutic in one patient who received FOLFOX4. The observed heterogeneous patterns at baseline and during dose optimization of 5-FU levels suggest variations in 5-FU metabolism among treated patients. Physiological and/or genetic differences underlying heterogeneity in 5-FU levels during dose optimization require further study of patient demographics, single nucleotide polymorphisms in Dihydropyrimidine Dehydrogenase (DPD), TS, or other genes that impact 5-FU metabolism and gene expression changes in liver after 5-FU therapy.