The goal of the research is to explore and analyze fuzzy classification algorithms, determine their advantages over classical algorithms and give recommendations for practical use of the algorithms. The inability of humans to understand computers and any other programmable hardware in linguistic terms has been one of the prerequisites for this trend to be so perspective. The most data available in real life and nature belong only partly to any predefined class, e.g. rainy weather belongs to the class ‘good weather’ with probability 0,3. This way classical data that have easily definable membership (belongs or does not belong) are rare. Therefore algorithms that can classify fuzzy data are becoming more necessary and irreplaceable in various scopes of life where there is a need to understand machinery using linguistic terms. The research uses data mining, mathematical logic and information theory methods, fuzzy logic, technique for extracting modular rules and automatic construction of membership functions. Classical PRISM and fuzzy PRISM algorithms were explored within this work to determine their differences and possibilities to improve results using fuzzy techniques. Different fuzzy algorithms were compared with each other – fuzzy PRISM and Finding relevant attributes and membership functions algorithm, to determine working principles of fuzzy classification algorithms. Other related methods that classify fuzzy data were also studied. The suitability of each algorithm for explored data sets was defined in order to achieve better results. Experiments were carried out to initially determine requirements towards data sets, then a search for suitable data sets was conducted and the most perspective ones were chosen for further experiments to select the most suitable data sets for use in a comparative study of possibilities of the algorithms. A plan of experiments was made for the most perspective data sets to perform a comparative analysis of fuzzy classification algorithms. The research gives recommendations for possibilities of use of the algorithms, depending on composition and structure of a data set. Conclusions of the work give practical information about the use of these algorithms according to features of a data set and outline possible directions of future research.