The aim of this study was to evaluate the performance of shade-determining devices. For the daily practitioner, it is essential to know whether modern computer-aided shade selection is reliable in everyday life. So the question of how the clinical usability of these machines could be rated has to be clarified. In the following, three actual devices available in the market were compared using a human observer's perception. The SpectroShade device (MHT Optic Research AG, 8155 Niederhasli, Switzerland), the ShadeVision device (X-Rite Co., Grandville, USA) and the Digital Shade Guide DSG4 (A. Rieth, 73614 Schorndorf, Germany) were assessed with respect to their agreement with the color perception of three examiners looking at 57 test persons (six teeth each for a total of 342). Shades were reported in Vita Classical shades. It could be demonstrated that every single human examiner showed a significantly higher agreement value (human group on average 40.2%) when compared with the remaining five methods than each computer-aided tooth shade determination device. The devices reached on average only a value of 28.6%, whereas the X-Rite ShadeVision showed a significant better result (33.2%) than the MHT SpectroShade and Rieth DSG4 (27.0 and 25.7%). Identical shade results given by all three methods of a group (group of three devices and three humans) were found to be rather low for the computer-aided devices (9.9%) compared with humans (36.7%). All six methods together agreed in 3.3% of the cases. It becomes evident that the methods-especially the computer-aided shade determination-are rather divided about the respective tooth color. Deficiencies of the instrumental as well as the visual detection become obvious. The best agreement level was performed by the human examiners. The best agreement of the evaluated devices was obtained-generally as well as among the human testers-by the X-Rite ShadeVision system, followed at a statistically significant distance by the MHT SpectroShade and the Rieth DSG4. The agreement among the examiner group was 52.9%, significantly better than that of each device compared to this group (31.3% on average). Color detection and its realization are very complex. As shown, in many cases, computer-aided color shade determination of natural teeth seems to not reflect human perception.