Since the 1980s Artificial Neural Networks (ANNs) have experienced a dramatic resurgence and have been applied to a broad range of different problems, mainly in engineering, medicine, finance and management. They have also become increasingly common in Nuclear Fusion, where the most interesting results have been obtained in the fields of signal processing, real time control and data analysis, where they can compete successfully or complement effectively traditional algorithmic approaches. With regard to signal processing, ANNs can be very useful in modelling non-linear transfer functions and solving ill-posed problems, in various forms of pattern recognition and identification. They have indeed been applied successfully in deriving global information from tomographic diagnostics and in identifying spectral lines for erosion monitoring. So far ANNs have probably found the most common applications in real time control, for their speed and generalisation capabilities. The identification of locked mode position and the estimate of the magnetic axis location are some examples in which their potentials have been decisive in obtaining good results within the required time. In the field of real time ANNs can also been used to complement algorithmic approaches, like the evaluation of the internal inductance estimated from the external magnetic measurements. As far as data analysis is concerned, ANNs can be particularly efficient in automatically extracting diffuse information from complex data, reducing the need of time consuming human supervision. A typical example is the derivation of relevant physical quantities from Langmuir probes. They can also be very useful in identifying the relevant parameters in complicated physical problems, implementing various forms of correlation techniques. The positive results obtained so far motivate the study of more complex architectures and the implementation of advanced concepts like fuzzy logic and adaptive control.