In psychopathological diagnosis, a correct classification of mental retardation level is needed to choose the best treatment for rehabilitation and to assure a quality of life suitable for the patient's specific condition. In order to meet this need this paper presents a new approach that permits performing automatic diagnoses efficiently and reliably, and at the same time is an easy-to-use tool for psychotherapists. The approach is based on a computational intelligence technique that integrates fuzzy logic and genetic algorithms in order to learn from samples a transparent fuzzy rule based on a diagnostic system. Empirical tests on a database of patients with mental retardation and comparisons with established techniques showed the efficiency of the proposed approach, which also gives a great deal of useful information for diagnostic purposes.