Limitations on the Accuracy Possible in Astrometric Observations of Satellites of the Major Planets Derek Jones, email:dhpj@ast.cam.ac.uk The limitations discussed include a) the accuracy with which the image of an object can be located on the detector; b) the accuracy and availability of stars of known position to act as reference points for measurement; and c) the accuracy with which positions on the detector can be mapped onto stars of known position; in particular the effect of atmospheric dispersion. One approach is to make auxiliary exposures of star clusters containing many stars of accurately known position (e.g. M15, Le Campion et.al.,1992, A & A Supp., 95,233) and to use the scale and orientation derived to reduce the exposures of planetary satellites. With a modern telescope with low-expansion optics there should be no problem with the scale. The detector will rotate relative to the celestial co-ordinate system because of the inevitable misalignment of the telescope mounting. With a computer-operated telecope the misalignment of the telescope mounting must be known to point the telescope and it is straightforward to adapt the telescope pointing model to predict the rotation of the detector. Another difficulty arises from the paucity of fields with sufficient stars of accurately known position. It is sometimes necessary to set up new fields using positions from wide-field photographic plates. The accuracy achieved in current work is compared to that theoretically attainable.