Measurement systems distort the time domain waveforms because of their limited bandwidth. The compensation of this distortion is called deconvolution. The aim of this project was to provide a support in the choice of the deconvolution algorithm, for different applications, different categories of transient signals, measurement systems and disturbing effects. We examined the methods of handling the uncertainty of the measurement and of the knowledge in the area of artificial intelligence. We developed the prototype of an expert system, based on fuzzy rules, which gives aids the user to chose the appropriate algorithm in a given deconvolution problem. The expert system was tested for many problems in the area of measurement technique. and the solutions provided by the expert system was in accordance with the choice of an expert of the area. The advantage of the system is that beside of providing the optimal algorithm a measure of applicability of all algorithms can be extracted, which helps the user to verify the decision.