THE QUESTION OF WHICH BIRTH COHORTS TO SELECT FOR LONGITUDINAL ANALYSES WAS ANSWERED BY DATA AVAILABILITY; SCHOOL, POLICE, AND COURT RECORDS WERE WELL ESTABLISHED FOR THE 1942, 1949, AND 1955 COHORTS. THE CUT-OFF DATE FOR CODING OF ALL RECORDS WAS 1977. THE 1942 COHORT CONSISTED OF 1,352 PERSONS, THE 1949 COHORT NUMBERED 2,099, AND THE 1955 COHORT DATA WERE OBTAINED FOR 2,676 PERSONS. SINCE MOST OF THE INDIVIDUALS STUDIED WERE WHITE, ONLY LIMITED ANALYSIS OF RACE AND ETHNIC DIFFERENCES WAS POSSIBLE. REASONS FOR POLICE CONTACT WERE INITIALLY CODED INTO 26 CATEGORIES FROM THE FILES OF THE RECORDS DIVISION OF THE RACINE POLICE DEPARTMENT. ADDITIONAL INFORMATION WAS OBTAINED THROUGH INTERVIEWS WITH COHORT MEMBERS. DATA ANALYSES SHOWED THAT PERSONS IN THE 1955 HAD MORE CONTACTS AND OF GREATER SEVERITY WITH THE POLICE THAN DID THOSE IN THE 1949 AND 1942 COHORTS. PERSONS IN THE 1949 COHORT HAD A HIGHER CONCENTRATION OF CONTACTS AMONG A SMALLER PROPORTION OF THE COHORT THAN DOES THE 1942 COHORT. THE RATE OF REFERRAL FOR CONTACTS HAD DECREASED SLIGHTLY DURING THE PERIOD OF EXPOSURE IN RACINE FOR THE 1942 AND 1949 COHORTS BUT INCREASED FOR THE 1955 COHORT. IN ADDITION, CONTINUITY IN CAREERS IS GREATER FOR THE 1949 COHORT THAN FOR THE 1942 COHORT. PREVIOUS CONTACTS AND SERIOUSNESS OF CONTACTS AS PREDICTORS OF FUTURE CONTACTS AND SERIOUSNESS WERE MORE EFFECIENT PREDICTORS FOR THE 1942 COHORT THAN FOR THE 1949 COHORT. WHEN MEASURES OF THE PAST WERE UTILIZED AS PREDICTORS, THERE WAS LITTLE OR NO INCREASE IN PREDICTIVE EFFICIENCY OVER THAT OBTAINABLE FROM THE MARGINALS. EXTENSIVE TABULAR STATISTICAL DATA ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED). (LWM)