Abstract: Vertical temperature profiles are inferred by a neural network based inverse procedure from satellite data. A multilayer perceptrons artificial neural network is trained using data provided by the direct model characterized by the Radiative Transfer Equation (RTE). Analysis of the neural network results reveals the obtained profiles are similar to the results of experimental data, showing a good performance of neural network based models for solving the inverse problem of moisture retrieval from satellite data. The advantages of using neural network systems are related to their intrinsic features of parallelism and its hardware implementation.