Abstract: Compared to classical linear and planar arrays, conformal array antennas possess some potential advantages including reduction of aerodynamic drag, wide angle coverage, space-saving, potential increase in available aperture, reduction of radar cross section, elimination of radome-induced bore-sight error and so on. They will find their promising applications in a variety of fields ranging from space-borne, airborne, ship-borne, missile-borne radar, space vehicles and wireless communications to sonar, etc. However, conformal array antennas integrated onto aircrafts and some other supporting structures are usually subjected to static deformations and vibrations caused by inertial forces and aerodynamic loads. Furthermore, its position errors and orientations of the elements are time varying due to the platform’s mechanical vibrations. As a result the performance of high-resolution direction finding algorithms such as MUSIC degrades severely. So this makes it essential to design robust DOA estimation algorithm and position errors calibration algorithm taking into account the influence of deformations and vibrations for such kind of conformal array antennas. This paper presents a robust direction finding algorithm and position errors calibration algorithm for conformal array antennas with time varying position errors. To obtain the modified time varying steering vector, mathematical model of vibrations for conformal linear array is given beforehand. The MUSIC spectrum is calculated with the time varying steering vector within a vibrating periodicity. By searching the peaks of the two dimensional (2-D) MUSIC spectrum, the ambiguous initial value of time in the mathematical model of vibration are estimated together with the DOAs. To resolve the ambiguity of initial value of time estimation, repeat the previous process with the snapshots data sampled after a very short time interval. Accordingly the time varying position errors of the elements can be predicted with the mathematical model of vibration and calibrated as well. Computer simulation demonstrates that the proposed algorithm is effective.