WHILE IT IS NOW GENERALLY ACCEPTED THAT CRIMINAL OFFENDERS DO NOT FORM A SINGLE HETEROGENEOUS GROUP, A CLASSIFICATION SYSTEM OF RELATIVELY HOMOGENEOUS BEHAVIORAL/PERSONALITY SUBGROUPS COULD PROVIDE USEFUL DATA IN DETERMINING APPROPRIATE TREATMENT OR REHABILITATION PROGRAMS IN APPLIED CRIMINAL JUSTICE SYSTEMS. FOR A TAXONOMIC SYSTEM TO BE USEFUL, IT MUST BE COMPLETE, RELIABLE, VALID, DYNAMIC, ECONOMICAL, AND CONTAIN CLEAR OPERATIONAL DEFINITIONS AND IMPLICATIONS FOR TREATMENT. NO PRESENT SYSTEM MEETS ALL THESE CRITERIA. SYSTEMS BASED ON THE CRIMINAL OFFENSE ARE UNRELIABLE BECAUSE DIFFERENT TYPES OF OFFENDERS MAY SHARE CERTAIN CHARACTERISTICS, THE OFFENSE MAY BE ATYPICAL, OR THE OFFENSE FOR WHICH AN INDIVIDUAL IS COMMITTED MAY BE THE RESULT OF PLEA BARGAINING. CRIMINAL CAREER CLASSIFICATIONS ARE APPLICABLE TO ONLY A FRACTION OF THE OFFENDER POPULATION. PHYSICAL OR ANTHROPOMETRIC CLASSIFICATION SYSTEMS HAVE PROVED UNFRUITFUL AND HAVE BEEN DISCARDED. OTHER SYSTEMS HAVE EMPLOYED DEMOGRAPHIC CHARACTERISTICS FOR CLASSIFICATION. ALL SUCH SYSTEMS ARE STATIC AND CANNOT REFLECT CHANGE OVER THE COURSE OF INCARCERATION. SYSTEMS USING PSYCHIATRIC DIAGNOSES ARE LIMITED IN APPLICABILITY AS MOST OFFENDERS DO NOT SUFFER FROM SPECIFIC PSYCHIATRIC DISORDERS, AND THE RELIABILITY OF DIAGNOSTIC CATEGORIES HAS BEEN QUESTIONED. SYSTEMS BASED ON BEHAVIORAL PATTERNS ARE EXPENSIVE, TIME CONSUMING, AND DEPEND ON THE RELIABILITY OF THE OBSERVERS. THE WARREN I-LEVEL SYSTEM AND THE QUAY SYSTEM RELY ON PERSONALITY ASSESSMENT. HOWEVER, BOTH OF THESE SYSTEM REQUIRE EXTENSIVE RATER TRAINING, ARE INEFFICIENT FOR MASS-SCREENING, AND ARE SUBJECT TO RATER ERROR. RESEARCH REPORTED IN SUCCEEDING ARTICLES (SEE NCJ 47697-47700) EXAMINES THE UTILITY OF THE MMPI, A STANDARDIZED PAPER-AND-PENCIL PERSONALITY INVENTORY, AS THE BASIS FOR A CLASSIFICATION SYSTEM WHICH WOULD MEET THE CRITERIA FOR PRACTICAL CRIMINAL JUSTICE APPLICATION. A BRIEF SUMMARY OF THE MMPI RESEARCH TO BE REPORTED IS INCLUDED. REFERENCES ARE PROVIDED. (JAP)