Certain natural fatty acids are taken up avidly by tumors for use as biochemical precursors and energy sources. We tested in mice the hypothesis that the conjugation of docosahexaenoic acid (DHA), a natural fatty acid, and an anticancer drug would create a new chemical entity that would target tumors and reduce toxicity to normal tissues. We synthesized DHA-paclitaxel, a 2'-O-acyl conjugate of the natural fatty acid DHA and paclitaxel. The data show that the conjugate possesses increased antitumor activity in mice when compared with paclitaxel. For example, paclitaxel at its optimum dose (20 mg/kg) caused neither complete nor partial regressions in any of 10 mice in a Madison 109 (M109) s.c. lung tumor model, whereas DHA-paclitaxel caused complete regressions that were sustained for 60 days in 4 of 10 mice at 60 mg/kg, 9 of 10 mice at 90 mg/kg, and 10 of 10 mice at the optimum dose of 120 mg/kg. The drug seems to be inactive as a cytotoxic agent until metabolized by cells to an active form. The conjugate is less toxic than paclitaxel, so that 4.4-fold higher molar doses can be delivered to mice. DHA-paclitaxel in rats has a 74-fold lower volume of distribution and a 94-fold lower clearance rate than paclitaxel, suggesting that the drug is primarily confined to the plasma compartment. DHA-paclitaxel is stable in plasma, and high concentrations are maintained in mouse plasma for long times. Tumor targeting of the conjugate was demonstrated by pharmacokinetic studies in M109 tumor-bearing mice, indicating an area under the drug concentration-time curve of DHA-paclitaxel in tumors that is 8-fold higher than paclitaxel at equimolar doses and 57-fold higher at equitoxic doses. At equimolar doses, the tumor area under the drug concentration-time curve of paclitaxel derived from i.v. DHA-paclitaxel is 6-fold higher than for paclitaxel derived from i.v. paclitaxel. Even at 2 weeks after treatment, 700 nM paclitaxel remains in the tumors after DHA-paclitaxel treatment. Low concentrations of DHA-paclitaxel or paclitaxel derived from DHA-paclitaxel accumulate in gastrocnemius muscle; which may be related to the finding that paclitaxel at 20 mg/kg caused hind limb paralysis in nude mice, whereas DHA-paclitaxel caused none, even at doses of 90 or 120 mg/kg. The dose-limiting toxicity in rats is myelosuppression, and, as in the mouse, little DHA-paclitaxel is converted to paclitaxel in plasma. Because DHA-paclitaxel remains in tumors for long times at high concentrations and is slowly converted to cytotoxic paclitaxel, DHA-paclitaxel may kill those slowly cycling or residual tumor cells that eventually come into cycle.