A retrodialysis (RD) method for the real-time calibration of on-line microdialysis (MD) procedures was investigated in vitro and in vivo. Calibration by retrodialysis was simultaneously validated through the use of a zero-net flux (ZNF) method, which assumes directional independence of diffusion of the solute. In RD, a calibrator with dialysance (PeA; effective permeability–surface area product) similar to that of the compound of interest is introduced into the perfusate. If the calibrator is suitable, its loss from the perfusate during RD is identical to the recovery of the solute of interest determined simultaneously by normal MD. Two antiviral nucleosides (AZT and AZdU) which differ structurally by only a methylene group were utilized as solute and calibrator, respectively. Both nucleosides exhibited similar recovery and loss at flow rates of 0.5 to 5 (µL/min in vitro, indicating a similar PeA product in this flow domain. Furthermore, both compounds showed similar loss into the lateral ventricle or thalamus of rabbits (n = 4) during RD at a flow rate of 1 µL/min for 6 hr. The relative loss decreased rapidly within the first hour, reaching a relatively stable value after 2 hr. The significant reduction in the loss of AZdU and AZT in vivo compared with that in vitro likely results from a lower diffusion coefficient in tissue. The distribution of AZT between plasma and cerebrospinal fluid (CSF) in the ventricle and extracellular fluid (ECF) in thalamus was determined at steady state using calibration by RD and ZNF simultaneously. The relative loss of AZdU during continuous RD was not significantly different from the recovery of AZT determined by ZNF in the same animal. Since RD may allow for continuous monitoring of microdialysis recovery in real time, it may offer an advantage over the ZNF method of system calibration. The steady-state Ccsf/Cp and Cecf/Cp ratios for AZT in this study were 0.26 ± 0.08 and 0.18 ± 0.08. That these ratios are much less than unity suggests that carrier-mediated transport of AZT exists in the brain-to-plasma direction.