Abstract: Water vapor plays a key role in atmospheric processes. Water vapor field is also one of the initial conditions needed by numerical weather prediction. Its distribution remains difficult to quantify due to water vapor's high variability in time and space and the deficiency of available measurements. The GPS has proved its capacity to measure the integ rated water vapor at zenith with the same accuracy as other methods, such as radiosonde and water vapor radiometers, which has been demonstrated by many experiments and GPS precipitable water vapor has been operated in many areas. Recent studies also show that it is possible to quantify the integ rated water vapor in the direction of signals of the GPS satellite. These observations can be used to study the water vapor tomography using GPS network. That is to use the water vapor amount along slant path of GPS satellites to form observation equations and to obtain the vertical profile information over GPS sites by solving these equations.The principle of tomography is introduced in which slant path observations from ground-based GPS is used and the method of calculating the observation equations is presented. During November in 2005, a small GPS network experiment is carried out in Hainan. The purpose of this experiment is to study the tomography tropospheric water vapor profiles based on GPS slant path observations. Without a priori information of water vapor, the absolute value of water vapor is difficult to resolve. After the method of GPS raw data processing is summarized, how to use a priori information is also analyzed. The period of tomography is from Nov 17 to 21 in 2005. Three different vertical tomographic resolutions of 600 m, 800 m and 1000 m are tested and analyzed. Results show that these resolutions have no large difference. Therefore, tomographic resolution from 500 m to 1000 m can be applied in practical use. In the experiment, tomographic vertical resolution of 1000 m is selected. Three different plans using a priori information are described.In the first plan, priori information is the earliest radiosonde at 00:00 on Nov 17, 2005. In the second plan, the latest radiosonde before the retrieved time is used as priori information. The average radiosondes are used in the third plan during the experiment. In the first and the third plan, priori information is hold but in the second plan priori information is changed and regenerated. Water vapor vertical profiles derived from GPS tomography are compared with that measured by radiosonde in Haikou site. The results show that they agree well with each other and the root-mean-square error (RMSE) is about 0.5 g·m ^-3. The process of water vapor decreasing and atmosphere drying is well reflected by GPS tomography. The tomographic results from different priori information used in tomographic method are also analyzed. It is demonstrated that water vapor priori information is obviously adjusted by GPS observation. And some problems in tomography are also discussed. Vertical profile distribution of tropospheric water vapor can be retrieved from GPS measurements with high time resolution through tomography inversion.