Abstract: In order to meet the needs of flood control and drought relief, the operation of soil moisture observation is launched routinely in meteorological department, by artificial boring stick all the time or by automatic measurer in recent years. However, the use of soil moisture data is always lagging with poor matching service and continuity. Based on the soil water observation network (including the manual and automatic network) and many kinds of approaches for data transmission, Real Time Observing and Forecasting System for Soil Moisture in Anhui Province (SMRTOFS) is developed. SMRTOFS is composed of data observation and transmission subsystem, forecast subsystem, and display subsystem. In data observation and transmission subsystem, the data from manual observers and automatic observation stations is collected in real time and stored in standard soil moisture database, and the data from unexpected transmission approach is also automatically gathered and conserved by defining an intermediate file. In forecast subsystem, predicting models of soil water content for each season are established, and soil moisture forecast is achieved using the latest soil water observation data and the coming 10-day weather information. In the display subsystem, based on the secondary development of Golden Software Surfer 8.0 and line bar chart control, the results of soil water observation and prediction in different seasons and different depths are exported and displayed dynamically, with the patterns of data table, the filled contour in spatial scale, bar chart, and so on. In the system, four-level files from observation to application are constructed including observation raw data, standard database, primary products and user products. The operation flow of soil moisture observation and forecast is reduced to transforming the four-level files. With higher applicability and compatibility, the system is applied triumphantly to the service of agricultural drought and waterlogging operation in Anhui Province. The information could be used to avoid the loss of flood and drought disaster. However, the soil moisture forecasting is based on statistical method, so the model parameters need modification for other regions. Implementing better Soil-Plant-Atmosphere Continuum model can also improve the performance of this system.