The Glueck project, which was undertaken during the 1940's, involved a matched sample of 500 officially designated delinquents and 500 nondelinquents. Data were collected on the 500 delinquents and 500 nondelinquents over 25 years in three waves: at the first interview (average age 14), at the subject's 25th birthday, and at the subject's 32nd birthday. Information was collected on social, psychological, biological, developmental, family, employment, school performance, and life events. Data used in the current study were the individual criminal histories as events of 480 delinquent boys representing their arrests from birth to 32 years old. Offense specialization was found at some level for virtually all crime types within all of the subgroups. The strongest level of juvenile specialization was for auto offending, and this was also the case for persistent juvenile offenders. Violent offending was significantly specialized for juveniles as a group and higher for persistent offenders. For adults there was significant specialization for all categories except violent offending. The level of specialization in burglary was higher for adults than for juveniles. For juveniles as a group, the overall direction of offending did suggest de- escalation. For persistent juvenile offenders, there was an overall trend toward escalation in the seriousness of offending over time. Regarding policy implications of the findings, the only group that would seem to be a candidate for selective incapacitation would be the juvenile persistent offender; however, by the time persistent juvenile offenders are identified, a significant amount of their criminal offending has already occurred. Identification based on characteristics other than offending patterns could address this problem. Theoretical implications and implications for future research are also drawn. Extensive tables and 80 references