Reconstituted multilayered oesophageal epithelium appears to be a good basis to test the efficacy of voriconazole (VOR) and fluconazole (FLU) in the tissue. The resulting model of a Candida oesophagitis was approaching the in vivo situation. We infected the tissue with 2 x 10(6) cfu of the Candida albicans strain SC5314. In the trials with FLU we also used clinical strains. Four hours after infection a good growth of C. albicans appeared mainly with hyphae on the surface of the tissue and a tendency to invasion. The destruction of the tissue began after 36 h. VOR (2 and 16 microg ml-1, respectively) prevented the penetration of hyphae into the tissue, when it was given 4-8 h after infection. It was less effective in reduction of Candida growth on the tissue surface. When VOR was given 16-24 h postinfection, the Candida infiltration stopped more slowly. Thirty-six hours after infection VOR application could not stop the destruction of the tissue despite reducing the fungi. The results with FLU (32 microg ml-1) were in principle the same, but not so distinct. FLU seems to be more effective against clinical strains of C. albicans than against the type strain.