The results of the characterization of precipitable water vapor in the atmospheric column carried out in the context of identifying potential sites for the deployment of the Thirty Meter Telescope (TMT) are pre- sented. Prior to starting the dedicated field campaign to look for a suitable site for the TMT, candidate sites were selected based on a climatology report utilizing satellite data that considered water vapor as one of the study vari- ables. These candidate sites are all of tropical or subtropical location at geographic areas dominated by high-pressure systems. The results of the detailed on-site study, spanning a period of 4 yr, from early 2004 until the end of 2007, confirmed the global mean statistics provided in the previous reports based on satellite data, and also confirmed that all the candidate sites are exceptionally good for astronomy research. At the locations of these sites, the atmospheric conditions are such that the higher the elevation of the site, the drier it gets. However, the data analysis shows that during winter, San Pedro Martir, a site about 230 m lower in elevation than Armazones, is drier than the Armazones site. This finding is attributed to the fact that Earth's atmosphere is largely unsaturated, leaving room for regional variability; it is useful in illustrating the relevance of in situ atmospheric studies for understanding the global and seasonal variability of potential sites for astronomy research. The results also show that winter and spring are the driest seasons at all of the tested sites, with Mauna Kea (in the northern hemisphere) and Tolonchar (in the southern hemisphere) being the tested sites with the lowest precipitable water vapor in the atmospheric column and the highest atmospheric transmission in the near and mid-infrared bands. This is the tenth article in a series discussing the TMT site-testing project.