Immunotoxins have been suggested as possible therapeutic agents in patients with leptomeningeal carcinomatosis. The pharmacokinetics, stability, and toxicity of immunotoxins injected into the i.t. space were examined in rats and rhesus monkeys. Monoclonal antibodies specific for the human (454A12 and J1) and rat (OX26) transferrin receptors were coupled to recombinant ricin A chain. In monkeys, the maximally tolerated dose of the anti-human transferrin receptor immunotoxin (454A12-rRA) was a dose that yielded a nominal cerebrospinal fluid (CSF) concentration of approximately 1.2 x 10(-7) M. In rats, the 10% lethal dose (LD10) of the anti-human transferrin receptor immunotoxin was a dose yielding a nominal CSF concentration of 8.8 x 10(-7) M whereas the LD10 of the anti-rat transferrin receptor immunotoxin (OX26-rRA) was a dose yielding a nominal CSF concentration of 1.2 x 10(-7) M. Thus, the species-relevant antibody resulted in toxicity at a concentration one-seventh that of the immunotoxin with the irrelevant antibody. A comparison of the area under the concentration curve at the LD10 for rats with the area under the concentration curve at the maximally tolerated dose in monkeys and humans shows that the species-relevant immunotoxin was a better predictor of the toxic dose of the anti-transferrin receptor immunotoxin in humans than the irrelevant immunotoxin. The pharmacokinetics of the 454A12-rRA immunotoxin within the CSF of monkeys showed a biphasic clearance with an early-phase half-life of 1.4 h and a late phase half-life of 10.9 h. The clearance was 4.4 ml/h or approximately twice the estimated clearance due to bulk flow of CSF. Loss by degradation was ruled out because immunoblot analysis showed that the immunotoxin was stable for up to 24 h after administration. Possible losses in addition to sampling include diffusion into brain tissue and transcapillary permeation. The apparent volume of distribution was 10.1 ml or approximately three-fourths the total CSF volume of the monkey. Dose limiting toxicity corresponded with the selective elimination of Purkinje cells in both rats and monkeys and was manifested clinically as ataxia and lack of coordination. The onset of ataxia in monkeys occurred within 5 days and, in the more mild form, was reversible with time. There was evidence of only minimal inflammation within the CSF, and there were no signs of systemic toxicity. Immunotoxins injected into the subarachnoid space may have potential for treatment of leptomeningeal carcinomatosis.