Red supergiants are massive, evolved stars that are among the brightest stars in the near infrared. But the uncertain physics of mass loss limits the ability of evolutionary models to accurately represent these stars in detail. Lacking a fundamental, predictive theory of mass loss, this process is normally incorporated into stellar models using simple parametric formulas such as Reimer’s Law. In this situation, it is important to constrain theoretical models evolved using such mass loss parametrizations by observation. A key stellar parameter is mass. In combination with a star’s position in the H-R diagram, a well-determined mass provides a strong constraint on stellar evolution models. Unfortunately, there are very few red supergiants with masses known to even 5% accuracy. In response, the objective of Hubble Space Telescope GO program 14070 (PI: Bennett) was to determine the mass of the red supergiant K star in the long-period eclipsing binary 31 Cygni (K4 Ib + B3 V) accurate to 1%. To obtain this accuracy requires the incorporation of new optical radial velocity observations of the supergiant; that work is in progress. But the observational program of GO-14070 is now complete, and these new HST/STIS observations of the 31 Cyg binary in the ultraviolet have provided an immediate determination of the secondary orbit radial velocity amplitude, K2, accurate to 1%. This result suffices to determine the masses of both 31 Cyg stars to an accuracy of about 2%, and provides a useful constraint on the evolution of the intermediate-mass red supergiant in this well-separated binary system.