The purpose of this study was to assess the factors affecting the calibration of the microdialysis probe for the in vitro recovery of triamcinolone acetonide (TA). Recoveries of TA were determined in microdialysis, retrodialysis, and no-net flux methods. Experiments were performed at room temperature or 37 degrees C while the reservoir medium was either stirred or unstirred. The effect of the viscosity of the medium on the recovery was studied using methylcellulose gel spiked with TA. Recovery was also calculated by the no-net-flux method in Ringer's solution and in plasma. Stirring the medium increased the recovery of TA by 30%. The recovery was higher at 37 degrees C under stirred or unstirred conditions and was same in either direction of dialysis. Increasing viscosity of the reservoir medium decreased the recovery (55% in Ringer's solution to 14% in 20% methylcellulose gel). Recovery from spiked plasma under stirred conditions was only 15% and this shift which was also seen in no-net-flux method was accounted for by the protein binding. Binding of TA, determined by ultrafiltration, was 20% in 5% gel and 81% in plasma. The recovery determined by the no-net-flux method was similar to the retrodialysis result. Stirring, temperature, viscosity and protein binding in the reservoir medium affected the in vitro recovery of TA.