By analyzing the amount of ligand remaining in the brain after microinjection into the brain cortex, the apparent efflux rate constants (Keff) of 3'-azido-3'-deoxythymidine (AZT) and 2',3'-dideoxyinosine (DDI) across the blood-brain barrier at low concentrations were determined to be 0.0317 +/- 0.0068 min(-1) and 0.0253 +/- 0.0037 min(-1), respectively. At higher concentrations, efflux exhibited saturation. The concentration of unlabeled DDI to inhibit 50% of the saturable efflux of [3H]DDI was found to be 11.3 +/- 5.7 microM, assuming that DDI diffused into the same volume of brain as that of trypan blue after intracerebral administration. The efflux rate of [3H]AZT from the brain was significantly inhibited by DDI, probenecid, p-aminohippuric acid, benzylpenicillin and 4,4'-diisothiocyanatostilbene-2,2'-disulfonic acid, but not by thymidine. Moreover, the efflux rate of [3H]DDI was significantly inhibited by AZT and probenecid, but not by deoxyinosine and inosine. After intracerebroventricular injection, the apparent efflux clearances of [3H]AZT and [3H]DDI from the cerebrospinal fluid were significantly inhibited by the coadministration of probenecid. However, intracerebroventricularly administered probenecid had no effect on the efflux of [3H]AZT and [3H]DDI from the brain after intracerebral microinjection, which suggested that the efflux transport system of the blood-cerebrospinal fluid barrier is not responsible for the elimination of AZT and DDI from the cerebral cortex. These results provide kinetic evidence that AZT and DDI are transported from brain into circulating blood across the blood-brain barrier via a probenecid-sensitive carrier-mediated efflux transport system.