This paper discusses the principles of adaptive optics, its performance, and its requirements for applications in astronomy to overcome limitations due to atmospheric turbulence. Guidelines for the implementation of these devices in telescopes are given in particular for the Very Large Telescope (VLT) of the European Southern Observatory. It is intended to equip each (me of the four 8-meter telescopes of the ESO- VLT, which are arranged in a linear array with an independent adaptive optical system. These systems will serve the individual and the combined Coude foci. In a first approach diffraction limited imaging for 3.5μm and longer is envisaged resulting a resolution of a few hundredths of an arcseconds. Currently, a small scale prototype adaptive system is under development. It is equipped with a 19 piezoelectric actuator deformable mirror, a Shack-Hartmann type wavefront sensor, and a dedicated wavefront computer for closing the feedback loop. This system is based on a polychromatic approach, i.e. it senses the wavefront in the visible but the adaptive correction loop works for the infrared wavelength range from .1 to 5 μm. The experience with this system will be used for the development of the final VLT systems, requiring a minimum of 150 to 200 subapertures. To solve the reference source problems experiments to generate artificial reference stars by scattering a laser pulse in the upper atmosphere are in preparation. Major developments are still necessary to solve the data processing problems which are associated with large numbers of subapertures and high correction bandwidth, especially if one plans later to extend the wavelength range towards the visible. In the VLT project adaptive optics is of particular importance for its synthetic aperture observation mode as a long baseline interferometer with resolutions in the range of a few milliarseconds on the sky. In this application a complete phasing of the telescope array is required in order to have the full gain of the large telescopes.