Spectra taken by the Mars Global Surveyor Thermal Emission Spectrometer (TES) have been used to monitor the latitude, longitude, and seasonal dependence of water vapor for over one full Martian year (March 1999 to March 2001). A maximum in water vapor abundance is observed at high latitudes during midsummer in both hemispheres, reaching a maximum value of ~100 pr-¿m in the north and ~50 pr-¿m in the south. Low water vapor abundance (<5 pr-¿m) is observed at middle and high latitudes in the fall and winter of both hemispheres. There are large differences in the hemispheric (north versus south) and seasonal (perihelion versus aphelion) behavior of water vapor. The latitudinal and seasonal dependence of the decay of the northern summer water vapor maximum implies cross-equatorial transport of water to the southern hemisphere, while there is little or no corresponding transport during the decay of the southern hemisphere summer maximum. The latitude-longitude dependence of annually averaged water vapor (corrected for topography) has a significant positive correlation with albedo and significant negative correlations with thermal inertia and surface pressure. Comparison of TES results with those retrieved from the Viking Orbiter Mars Atmospheric Water Detectors (MAWD) experiments <Jakosky and Farmer, 1982> shows some similar features but also many significant differences. The southern hemisphere maximum observed by TES was not observed by MAWD, and the large latitudinal gradient in annually averaged water vapor observed by MAWD does not appear in the TES results.