Microdialysis coupled to HPLC was used to study the disposition of local anesthetics in the cerebrospinal fluid (CSF) because of the difficulty in sampling CSF. A retrodialysis method for the microdialysis calibration was investigated in vitro and in vivo. Calibration by retrodialysis was simultaneously validated through the use of the zero net flux method. Two local anesthetics (bupivacaine and ropivacaine), which differ structurally by only one methyl group, were respectively utilized as substance of interest and as internal standard. Different parameters were tested in vitro to compare the relative recovery (RR) of bupivacaine and the relative loss (RL) of ropivacaine. Several flow rates were tried to select an optimal in vivo flow rate (1 microliter/min). the RR and RL values were not influenced by the variation of bupivacaine concentration. A significant variability among different probes within a batch was established (RR ranging from 41.1-65.3%; RL ranging from 30.7-61.0%). The K-factor values, defined as RLropivacaine/RLbupivacaine, were calculated in vitro and in vivo. This ratio decreased in vivo but was constant (K in vitro = 1.06 +/- 0.04, K in vivo = 0.87 +/- 0.03). The extracellular tissue concentration of the compound of interest was again in vitro and no deterioration of probe during the in vivo experiment was found. After administration of bupivacaine in the epidural space of rabbits, plasma and microdialysis CSF samples were simultaneously collected. Plasma and CSF disposition of bupivacaine displayed different kinetics. The maximum CSF concentration of B averaged 394 +/- 170 micrograms ml-1 with a mean Tmax of 3.8 +/- 1.8 min. The maximum CSF concentration of B averaged 0.44 +/- 0.09 micrograms ml-1 with a mean Tmax occurring at 1 min. Microdialysis, combined with accurate calibration, should be a reliable technique to gain further insight in the spinal disposition of local anesthetics.