The invention relates to a method for inverting atmospheric water vapor content from MODIS (Moderate Resolution Imaging Spectroradiometer) data, which can be applied to the remote sensing application departments, such as meteorology, environment monitoring, land management, crop condition monitoring, disaster monitoring and the like. The method comprises the following three steps of: (1) performing forward simulation to the second, fifth, seventeenth, eighteenth and nineteenth near-infrared bands of acquired remote sensing MODIS data in given areas and seasons and an atmospheric model by using atmospheric radiation transmission simulation software MODTRAN, and establishing a training and test database; (2) repeatedly training and testing the training and test data set by using a neural network; and (3) performing inversion calculation to the MODIS actual image data, and performing actual earth surface verification and application analysis. The atmospheric water vapor inversed product has high precision, strong practicability and relatively simple operation. The method can be applied to the departments of weather forecast, environment monitoring, crop condition monitoring, disaster monitoring and the like.