We have examined Uranus' radio spectrum and the latitudinal variation in its radio brightness temperature. We conclude that Uranus' spectrum cannot be explained with models based upon thermochemical equilibrium. The spectrum can only be matched if there is a low ammonia volume mixing ratio (a few times 10 -7 between roughly 150 < T < 240 K), distributed uniformly over a large range of altitude. To remove sufficient NH3 at deep tropospheric levels through the formation of an NH4SH-solid cloud, a large concentration of H2S gas is needed, which implies that sulfur must be enhanced relative to its solar value by a factor of a few hundred. We further can constrain the S/N ratio to be at least live times the solar value if the H20 volume mixing ratio is <100x solar and equal to or larger than 3x solar if H20 is >~500x solar. The elemental ratios derived from our work support the theory of planetary accretion by Pollack and Bodenheimer (1989, in Origin and Evolution of Atmospheres (S. Atreya, Ed.)). The constant mixing ratio of NH3 in the 145 < T ~< 240 K range implies either vertical mixing on time scales much shorter than expected from reasonable values of the eddy diffusion coefficient, or supersaturation of NH4SH at levels where T ~ 240 K. The pole-equator gradient in Uranus' radio brightness temperature implies a latitude-dependent variation in the ammonia mixing ratio, ranging from a few times 10 -6 at altitudes where the temperature is 145 < T < 240 K in the equatorial region (latitudes <30-40°), a few times 10 7 at 145 < T < 220 K at midlatitudes, and --10 -7 down to 280 K in the polar region. Whereas the values in the equator and midlatitudes can be explained by condensation theories, the polar value can only be explained by invoking strong downdrafts of dry air (gas from which most of the ammonia vapor has been removed by condensation). A comparison of images taken at 2-and 6-cm wavelength show a difference in the spatial distribution of the brightness temperature. In addition, images at both wavelengths show time variability in the zonal brightness distribution.