Knowledge of the exact dose and rate at which an antitumor agent is delivered to its target site is postulated to be crucial to proper patient management. It is now possible to obtain such information using non-invasive 19F-NMRS (nuclear magnetic resonance spectroscopy) following the administration of 5-fluorouracil (5-FU). We have performed such studies in 103 patients with breast, colorectal and other tumors. Measurable 19F signals were detected in 99 of these patients (92.5%). Estimation of the tumoral t1/2 of 5-FU in these patients revealed that 51 of them (51.5%) exhibited a tumoral t1/2 greater than 20 min, a value we had characterized as indicating drug trapping in the tumor. Of these patients, 46 who received regimen bolus 5-FU 600 mg/m2 with leucovorin for their treatment have been evaluated. In these patients, the association between trapping and response remains very high (p<.000001). None of the non-trappers responded to chemotherapy, whereas 70% of the evaluable trappers responded. Details are presented here on the methodology of NMRS data acquisition and on their pharmacokinetic analysis. The potential mechanisms underlying the trapping effect appear to be predicated primarily on transport processes. Suggestions are presented on how such pharmacokinetic imaging studies may extend both our understanding of the mechanism of action of 5-FU, and how they could be used to optimize patient treatment.