The objective of this study was to evaluate the performance of a new, compact, dynamic diffusion cell for in vitro transdermal permeation. These so-called Kelder-cells were developed as an automated alternative to the static Franz diffusion cells. The new cells were used in combination with the ASPEC-system (automatic sample preparation with extraction columns) which was initially designed for the automation of solid-phase extractions. Three variables were tested to optimize the performance of the new cell system: injection height into the inlet compartment, volume flowing through the receptor compartment and temperature. Experiments were performed using the tritium labelled anticholinergic [3H]dexetimide permeating through an artificial membrane (Silastic). The injection height of the needle into the inlet compartment of the cell should be programmed at -34 mm to ensure complete air tightness, thus forcing the buffer to flow through the cell. The volume of buffer flow through the receptor compartment is important in maintaining sink conditions: a volume of 117 microliters was chosen to replace the total content of the cell (84 microliters) every 2 min. The temperature was precisely controlled in a thermostatic cabinet to minimize variations in experimental conditions. For [3H]dexetimide, an increase in temperature of 20 degrees C reduced the lag time by a factor of approximately two, however the influence on the flux was negligible. The data for the Kelder-cells were comparable with static Franz diffusion cells at a pseudo-steady state, however Kelder-cells have the advantage of automatic sampling, continuous replacement of the receptor solution, and unattended operation over at least 24 h.