The dataset used in the study consisted of all cases referred to the juvenile court of Missouri in 1997. In order to measure offense severity in these cases, an ordinal variable called offense severity was created. This was done by examining the major allegation that was the basis for the referral to juvenile court. All cases that involved juveniles who were referred to the court for status offenses or misdemeanors against people or property were designated as minor (nonserious) and were coded with a zero on the offense severity variable (n=48,233). Cases that involved juveniles accused of violent felonies (physical victimization of a person) were designated as serious and were coded with a one on the offense severity variable (n=2,629). Juveniles who were referred for other types of offenses, such as drug offenses and felony property offenses were removed from the dataset (n=13,604), leaving 50,862 cases for analysis. Almost 73 percent of these cases involved White juveniles, and cases that involved boys outnumbered those that involved girls by a similarly wide (66.43 percent) margin. Black male juveniles were disproportionately referred for serious offending, and White girls were disproportionately referred for nonserious offenses. Black girls and White boys were referred for serious and nonserious offenses in almost the exact proportions that would be expected based upon their representation among the study cases. The researchers conclude that being Black significantly contributed to the likelihood of receiving detention, as it was the largest single contributor to the detention decision for both serious and nonserious offenders. One way of preventing juvenile justice decisionmakers from discriminating against girls and minorities in detention decisions is to develop guidelines that establish neutral criteria for detention, commitment, and other stages in the process that significantly curtail a juvenile's freedom. Another approach is to track race, gender, legal variables, and outcome measures on a yearly basis to determine whether particular judges are engaged in discriminatory decisionmaking. 5 tables, 10 notes, and 45 references