Context. Thanks to its excellent 5100 m high site in Chajnantor, the Atacama Pathfinder Experiment (APEX) systematically explores the southern sky at submillimeter wavelengths, in both continuum and spectral line emission. Studying continuum emission from interstellar dust is essential to locating the highest density regions in the interstellar medium, and deriving their masses, column densities, density structures, and large-scale morphologies. In particular, the early stages of (massive) star formation remain poorly understood, mainly because only small samples of high-mass proto-stellar or young stellar objects have been studied in detail so far. Aims. Our goal is to produce a large-scale, systematic database of massive pre-and proto-stellar clumps in the Galaxy, to understand how and under what conditions star formation takes place. Only a systematic survey of the Galactic Plane can provide the statistical basis for unbiased studies. A well characterized sample of Galactic star-forming sites will deliver an evolutionary sequence and a mass function of high-mass, star-forming clumps. This systematic survey at submillimeter wavelengths also represents a preparatory work for Herschel and ALMA. Methods. The APEX telescope is ideally located to observe the inner Milky Way. The Large APEX Bolometer Camera (LABOCA) is a 295-element bolometer array observing at 870 μm, with a beam size of 19. 2. Taking advantage of its large field of view (11. 4) and excellent sensitivity, we started an unbiased survey of the entire Galactic Plane accessible to APEX, with a typical noise level of 50−70 mJy/beam: the APEX Telescope Large Area Survey of the Galaxy (ATLASGAL). Results. As a first step, we covered ∼95 deg 2 of the Galactic Plane. These data reveal ∼6000 compact sources brighter than 0.25 Jy, or 63 sources per square degree, as well as extended structures, many of them filamentary. About two thirds of the compact sources have no bright infrared counterpart, and some of them are likely to correspond to the precursors of (high-mass) proto-stars or protoclusters. Other compact sources harbor hot cores, compact H ii regions, or young embedded clusters, thus tracing more evolved stages after massive stars have formed. Assuming a typical distance of 5 kpc, most sources are clumps smaller than 1 pc with masses from a few 10 to a few 100 M . In this first introductory paper, we show preliminary results from these ongoing observations, and discuss the mid-and long-term perspectives of the survey.