The theory of fuzzy classification with applications to geophysical data is explored. In the context of expert systems fuzzy classification helps improve the efficiency of the inference mechanism and the development of “fuzzy rules” for automatic classification. The proposed technique is motivated by the inaccuracy and unreliability that often exist in the geophysical data which adversely affect the performance of the Bayesian classifiers. Fuzzy classification, through its membership function, offers an elegant solution to the problem. The algorithm consists of supervised and unsupervised components. These two components interact in a hybrid fashion depending upon the level of uncertainty associated with the control samples. The algorithm takes into account the uncertainty in the data by assigning continuous membership grades to the samples with respect to the classes. Examples from seismic data illustrate the geophysical applications of the method.