BACKGROUND: The recommended thickness of the graft lamella in Descemet's stripping automated endothelial keratoplasty (DSAEK) is 120 - 180 microm. To adjust for the large variation in central corneal thickness (CCT) of eye bank donor corneas, microkeratome plates of different heights are available. The aim of this study was to evaluate the lamella thickness predictability of different keratome plates in a clinical routine setting. METHODS: In a prospective study of 60 consecutive DSAEK procedures, CCT was measured with a 50 MHz ultrasound pachymeter after meticulous epithelium removal immediately before and after lamella creation. Selected plate thickness was 300, 350, or 400 micro, respectively. Cutting depth was calculated by subtracting the remaining CCT of the deep lamella after the cut from the initial CCT before the cut. RESULTS: The mean (+/- SD) cutting depth was 320 +/- 45 microm with the Amadeus 350 keratome, 317 +/- 48 microm with the Moria 300 keratome, 388 +/- 58 microm with the Moria 350 keratome and 467 +/- 94 microm with the Moria 400 keratome. Neither the duration nor the type of donor preservation had an effect on the accuracy of the cutting depth. CONCLUSION: The mean cutting depth for the DSAEK keratomes is more accurate than the cutting depth of the same keratome heads designed for LASIK. This might be due to the fact that DSAEK lamellas were cut after complete epithelial removal. Despite the accurate mean cutting accuracy, there is a substantial variation between individual cuts which have to be taken into consideration.