The ASTROS Star Tracker (AST), designed and built by JPL as a demonstration instrument in the early 80s was integrated with the astro scientific payload in 1985 for a planned March 1986 launch. After a slip of nearly five years, it was finally launched in December 1990, gathering extensive scientific data for over one hundred scientific targets. This paper reviews some of the AST results from the Astro-1 flight. After the five-year stand down, photometric, spectral, and geometric calibrations remained virtually unchanged, allowing predictable performance on all targets and successful automatic identification of every star field. Although small changes in the optical point-spread function increased the centroid error, this did not affect operation for Astro, and should be correctable for future instruments. Our data suggest that calibration of centroid error to the 1/100 pixel level is achievable when the point-spread function remains stable. The data are also consistent with the noise-equivalent-angle (NEA) of 1/300 pixel measured in the laboratory for bright stars.