The usage of modern methods, which represent predictive neural techniques is an effective approach to the estimation of the processes occurring in the complex empirical systems of agricultural engineering. The artificial neural networks are a rapidly expanding field of knowledge used increasingly in many areas of science, as well as practice. The learning algorithms, enabling the design of appropriate network topology and selection of the parameters of this structure, matched to the problem to be solved are the basis of functioning of artificial neural networks. The paper proposes the use of neural modeling techniques to estimate the level of methane content in the biogas emitted over the methane fermentation process of silage. Obtained research results confirm the hypothesis that predictive neural model describing the methane production during the silage fermentation process in biofermentor is an appropriate tool to assess the forecasting of the level of this emission.