Long baseline optical/infrared interferometers, such as the Mark III Stellar Interferometer1 on Mt. Wilson and the ASEPS-0 Testbed Interferometer2 on Palomar Mountain, California, have good capabilities for narrow-angle and wide-angle astrometry with very high precision. Using the Mark III Interferometer many spectroscopic binaries became “visual” for the first time. The measurement accuracy of angular separation is 0.2 mas, the smallest separation measured between two components is 2 mas, the maximum magnitude difference is 4 mag, and the smallest semimajor axis is 4 mas. Such high angular resolution and dynamic range have been used to determine stellar masses with precision of 2% and differential stellar luminosities to better than 0.05 mag for separations of less than 0.″2. For some binary stars, not only have the systems been resolved, but also the diameter of the primary component has been determined, yielding direct measurements of stellar effective temperature with high accuracy. For parallax determination, the precision is 1 mas or better and is unaffected by interstellar extinction. For wide-angle astrometry with the Mark III interferometer, the observation results yielded average formal 1σ errors for FK5 stars of about 10 mas. Presently a new infrared interferometer, the ASEPS-0 Testbed Interferometer on Palomar Mountain is under construction, and is being optimized to perform high accuracy narrow-angle astrometry using long baseline observations at 2.2 μm, with phase referencing for increased sensitivity. The goal is to demonstrate differential astrometric accuracies of 0.06–0.1 mas3 in order to allow for detection of extra-solar planets in the near future.