DETENTION DECISIONS WERE STUDIED BECAUSE OF CONTROVERSY CONCERNING THE CRITERIA USED IN DECISIONMAKING AND THE POTENTIALLY HARMFUL EFFECTS OF DETENTION. THE TWO COURTS STUDIED, IN MEMPHIS AND IN DENVER, DIFFERED BOTH GEOGRAPHICALLY AND IN THEIR ORIENTATION TO JUVENILE JUSTICE. THE DENVER COURT WAS A DUE PROCESS JUVENILE TRIBUNAL, WHEREAS THE MEMPHIS COURT FOLLOWED THE MORE TRADITIONAL THERAPEUTIC MODEL. GOODMAN'S LOG LINEAR ANALYSIS METHOD UNCOVERED RELATIONSHIPS AMONG CATEGORICAL VARIABLES INCLUDING RACE, FAMILY INCOME, SEX, FAMILY STABILITY, PRIOR RECORD, PRESENT ACTIVITY AND TYPE OF COURT. IN DENVER, 24 PERCENT OF JUVENILES WERE DETAINED, COMPARED WITH 43 PERCENT IN MEMPHIS. RESULTS SHOWED THAT A JUVENILE'S PRESENT ACTIVITY AND PRIOR RECORD, AND THE COURT'S ATTITUDE TOWARD JUVENILE JUSTICE, WERE THE MAJOR DETERMINANTS OF DETENTION OR RELEASE PRIOR TO ADJUDICATION. IDLE YOUTHS AND THOSE WITH PRIOR OFFENSE RECORDS WERE MORE LIKELY TO BE DETAINED THAN THOSE IN CONVENTIONAL ACTIVITIES OR WITHOUT RECORDS. IN ADDITION, THE DETENTION DECISION WAS NOT BIASED BY THE JUVENILE'S RACE OR SOCIAL CLASS, BUT THERE DID APPEAR TO BE SEX BIAS, AS EXEMPLIFIED BY THE GREATER LIKELIHOOD OF DETAINING FEMALES REFERRED FOR DECORUM OFFENSES AND MALES REFERRED FOR PROPERTY OR VIOLENT OFFENSES. FURTHER RESEARCH ON THESE AND ADDITIONAL VARIABLES, AND USE OF MORE COMPLETE INFORMATION BY COURT PERSONNEL IN MAKING DECISIONS, IS RECOMMENDED. TABLES, ONE FIGURE, FOOTNOTES, AND A BIBLIOGRAPHY ARE INCLUDED. (CFW)