World-wide studies on bee diversity, conservation and on pollination ecology are hampered by the difficult taxonomy of bees, the lack of suitable literature and bee taxonomists. The automated identification system consists of an electronic notebook connected with a CCD camera mounted on a stereomicroscope. The identification of bees is based exclusively on characters of the fore-wing venation: The fore-wing is video-recorded and the image of the wing is transferred to the notebook. With a mouse-click the user marks defined vein junctions. The system then connects the junctions by automatic line-following and thus digitises the whole venation. The system has to be trained with a minimum of 30 well defined specimens of each sex per species. With data of each bee it learns and gets better. Species identification is achieved by automatic comparison of incoming data with already memorised data. Currently the system employs linear and non-linear discriminant analysis methods. We tested the system with very difficult cases like closely related species of Andrena, Bombus and Colletes which are a real problem for traditional taxonomy. In all cases the system identified the species with a confidence between 98 and 99,8%. This system can be applied by museum taxonomists as well as b y field workers with no special training in bee taxonomy. Dry specimens as well as live bees can be identified. Identification of a bee takes no more than 5 minutes. Wing images or readymade data can also be sent on disc or via internet to institutions which offer this automatic identification service.