Processing of aircraft surveillance data from several geographically separated radars is most easily accomplished using a common coordinate system to represent data from all sensors. The Multisensor Data Processing system currently being developed for the FAA in support of the Advanced Automation System (AAS) requires a degree of accuracy and consistency that is not available from the current NAS implementation of coordinate conversion. A study has been undertaken to design a coordinate covnersion algorithm that meets the needs of Multisensor Data Processing. The process of projection of the ellipsoidal surface of the earth onto a planar surface is examined in light of teh requirements of air traffic control systems. The effects of the non-spherical nature of the earth and of limited computational resources are considered. Several standard cartographic projection techniques are examined, and the sterographic projection is found to be the projection of choice. A specific implementation of stereographic projection that makes the needs of Multisensor Data Processing is described. This implementation makes use of several approximations to decrease the computational load. The systemic errors introduced by these approximations are removed by the addition of a correction term determined from a precomputed error surface. The performance of this conversion system is demonstrated using realistic test data.