The large uncertainties associated with measuring the amount of high temperature emission in solar active regions (ARs) represents a significant impediment to making progress on the coronal heating problem. Most current observations at temperatures of 3 MK and above are taken with broadband soft X-ray instruments. Such measurements have proven difficult to interpret unambiguously. Here, we present the first spectroscopic observations of the Fe XVIII 974.86 Å emission line in an on-disk AR taken with the SUMER instrument on SOHO. Fe XVIII has a peak formation temperature of 7.1 MK and provides important constraints on the amount of impulsive heating in the corona. Detailed evaluation of the spectra and comparison of the SUMER data with soft X-ray images from the X-Ray Telescope on Hinode confirm that this line is unblended. We also compare the spectroscopic data with observations from the Atmospheric Imaging Assembly (AIA) 94 Å channel on the Solar Dynamics Observatory. The AIA 94 Å channel also contains Fe XVIII, but is blended with emission formed at lower temperatures. We find that it is possible to remove the contaminating blends and form relatively pure Fe XVIII images that are consistent with the spectroscopic observations from SUMER. The observed spectra also contain the Ca XIV 943.63 Å line that, although a factor 2-6 weaker than the Fe XVIII 974.86 Å line, allows us to probe the plasma around 3.5 MK. The observed ratio between the two lines indicates (isothermal approximation) that most of the plasma in the brighter Fe XVIII AR loops is at temperatures between 3.5 and 4 MK.