The distribution of zidovudine (AZT) between plasma, brain extracellular fluid (ECF), and cerebrospinal fluid (CSF) was investigated in a crossover design study (n = 5) in unanesthetized rabbits. Drug was administered by intravenous (iv) and intracerebroventricular (icv) infusions at the same infusion rate (1.5 mg/h.kg). The concentrations of AZT in ECF and CSF were measured by HPLC with microdialysis sampling. Plasma concentrations of AZT were quantitated by HPLC. Following iv infusion, the ECF- and CSF-to-plasma concentration ratios at steady state (SS), were 0.19 +/- 0.05 and 0.29 +/- 0.06, respectively. These values were less than unity, indicating the existence of active transport processes for the transport of AZT from brain to plasma across the blood-brain barrier (BBB) or blood-CSF barrier (BCB). The transport processes were modeled by compartmental model analysis, and the results suggest that the transport efficiency of AZT across the BBB is asymmetric; that is, the efflux clearance was five times greater than the influx clearance. Similarly, the efflux clearance from CSF is three times larger than the influx clearance into CSF. The SS concentrations of AZT in brain ECF in the same animals that received an icv infusion of AZT in the crossover design study were approximately two orders of magnitude greater than those in animals following iv infusion at the same dosing rate. Nevertheless, the SS plasma concentrations of AZT were similar for both routes of administration (1.2 +/- 0.19 and 1.2 +/- 0.13 micrograms/mL for iv and icv routes, respectively), confirming that the brain is not an organ that exhibits first-pass metabolism under the present experimental conditions.