In animal models, liposomal formulations of paclitaxel possess lower toxicity and equal antitumor efficacy compared with the clinical formulation, Taxol. The goal of this study was to determine the formulation dependence of paclitaxel pharmacokinetics in rats, in order to test the hypothesis that altered biodistribution of paclitaxel modifies the exposure of critical normal tissues. Paclitaxel was administered intravenously in either multilamellar (MLV) liposomes composed of phosphatidylglycerol/phosphatidylcholine (L-pac) or in the Cremophor EL/ethanol vehicle used for the Taxol formulation (Cre-pac). The dose was 40 mg/kg, and the infusion time was 8 to 9 minutes. Animals were killed at various times, and pharmacokinetic parameters were determined from the blood and tissue distribution of paclitaxel. The area under the concentration vs time curve (AUC) for blood was similar for the 2 formulations (L-pac: 38.1 +/- 3.32 microg-h/mL; Cre-pac: 34.5 +/- 0.994 microg-h/mL), however, the AUC for various tissues was formulation-dependent. For bone marrow, skin, kidney, brain, adipose, and muscle tissue, the AUC was statistically higher for Cre-pac. For spleen, a tissue of the reticuloendothelial system that is important in the clearance of liposomes, the AUC was statistically higher for L-pac. Apparent tissue partition coefficients (K(p)) also were calculated. For bone marrow, a tissue in which paclitaxel exerts significant toxicity, K(p) was 5-fold greater for paclitaxel in Cre-pac. The data are consistent with paclitaxel release from circulating liposomes, but with efflux delayed sufficiently to retain drug to a greater extent in the central (blood) compartment and reduce penetration into peripheral tissues. These effects may contribute to the reduced toxicity of liposomal formulations of paclitaxel.