Current developments in tissue engineering and microtechnology fields have allowed the proposal of pertinent tools, microchips, to investigate in vitro toxicity. In the framework of the proposed REACH European directive and the 3R recommendations, the purpose of these microtools is to mimic organs in vitro to refine in vitro culture models and to ultimately reduce animal testing. The microchip consists of functional living cell microchambers interconnected by a microfluidic network that allows continuous cell feeding and waste removal controls by fluid microflow. To validate this approach, Madin Darby Canine Kidney (MDCK) cells were cultivated inside a polydimethylsiloxane microchip. To assess the cell proliferation and feeding, the number of inoculated cells varied from 5 to 10 x 10(5) cells/microchip (corresponding roughly to 2.5 to 5 x 10(5) cells/cm2) and from four flow rates 0, 10, 25, and 50 microL/min were tested. Morphological observations have shown successful cell attachment and proliferation inside the microchips. The best flow rate appears to be 10 microL/min with which the cell population was multiplied by about 2.2 +/- 0.1 after 4 days of culture, including 3 days of perfusion (in comparison to 1.7 +/- 0.2 at 25 microL/min). At 10 microL/min flow rate, maximal cell population reached about 2.1 +/- 0.2 x 10(6) (corresponding to 7 +/- 0.7 x 10(7) cells/cm(3)). The viability, assessed by trypan blue and lactate deshydrogenase measurements, was found to be above 90% in all experiments. At 10 microL/min, glucose monitoring indicated a cell consumption of 16 +/- 2 microg/h/10(6) cells, whereas the glutamine metabolism was demonstrated with the production of NH3 by the cells about 0.8 +/- 0.4 micromol/day/10(6) cells. Augmentation of the flow rate appeared to increase the glucose consumption and the NH3 production by about 1.5- to 2-fold, in agreement with the tendencies reported in the literature. As a basic chronic toxicity assessment in the microchips, 5 mM and 10 mM ammonium chloride loadings, supplemented in the culture media, at 0, 10, and 25 micaroL/min flow rates were performed. At 10 microL/min, a reduction of 35% of the growth ratio with 5 mM and of 50% at 10 mM was found, whereas at 25 microL/min, a reduction of 10% with 5 mM and of 30% at 10 mM was obtained. Ammonium chloride contributed to increase the glucose consumption and to reduce the NH3 production. The microchip advantages, high surface/volume ratio, and dynamic loadings, coupled with the concordance between the present and literature results dealing with ammonia/ammonium effects on MDCK illustrate the potential of our microchip for wider in vitro chronic toxicity investigations.