Recent calculations of atomic data for Fe XV have been used to generate theoretical line ratios involving n = 3-4 transitions in the soft X-ray spectral region (52-83 A), for a wide range of electron temperatures and densities applicable to solar and stellar coronal plasmas. A comparison of these with solar flare observations from a rocket-borne spectrograph (XSST) reveals generally good agreement between theory and experiment. In particular, the 82.76 A emission line in the XSST spectrum is identified, for the first time to our knowledge in an astrophysical source. Most of the Fe XV transitions which are blended have had the species responsible clearly identified, although there remain a few instances where this has not been possible. The line ratio calculations are also compared with a co-added spectrum of Capella obtained with the Chandra satellite, which is probably the highest signal-to-noise observation achieved for a stellar source in the 25-175 A soft X-ray region. Good agreement is found between theory and experiment, indicating that the Fe XV lines are reliably detected in Chandra spectra, and hence may be employed as diagnostics to determine the temperature and/or density of the emitting plasma. However the line blending in the Chandra data is such that individual emission lines are difficult to measure accurately, and fluxes may only be reliably determined via detailed profile fitting of the observations. The co-added Capella spectrum is made available to hopefully encourage further exploration of the soft X-ray region in astronomical sources.