A Philadelphia birth cohort study's finding that 6 percent of boys born there in 1945 experienced 52 percent of the arrests stimulated a variety of research and policy initiatives, including those relating to selective incapacitation. Results from longitudinal delinquency studies in London; Racine, Wis.; and Marion County, Oreg., paralell those of the Philadelphia study. A high percentage, typically a third, of cohort members were arrested. Many had only one or a small number of official contacts with the criminal justice system. A small percentage had 6 or more contacts and for these, the probability of subsequent recidivism after any contact was about 80 percent. The prospective identification of these chronic offenders could have significant impact on crime reduction efforts. Most incapcitation research, however, involved retrospective -- not prospective -- identification and has been characterized by high false positives. By contrast, the London study identified 7 variables that are apparent by age 10 (such as I.Q., family background, and school behavior problems) that may permit prospective identification of a substantial number of chronic offenders. The prediction results closely match the results of predictions based on a theoretical model that uses aggregate recidivism data to partition a cohort into innocents, desistors, and persisters. The results suggest the possibility of early identification between more and less serious offenders and also support the view that the rise in recidivism probability with increasing involvement in crime results from a changing mix of desisters and persisters among offenders. Provided are 7 tables, 4 figures, 18 references, and 20 footnotes. (Author abstract modified)