The purpose of this research was to test the relative ability of legal and extralegal variables to predict sentencing in a juvenile court. To this end, a sample of 464 cases of 14 and 15-year-old criminal offenders arrested in Newark, New Jersey in 1973 was studied. Multiple regression was used to examine the effect of the legal (e.g., seriousness of the offense, and previous arrest records) and extralegal (e.g., race, sex, and social class) variables, and only one legal factor, i.e., the number of previous arrests, had impact on sentencing decisions. However, the greatest amount of variance in judicial decisionmaking remained unexplained until, as followup to this research, the sample was followed for a 2-year period to find how the severity of the original disposition was related to the number of subsequent arrests. The analysis of these data showed that two variables--records of school and of family problems--had a significant relationship with the original disposition. These two social background variables would not be acceptable as decisionmaking criteria in adult court. However, in juvenile court, where a substantive, individualized justice, is practiced, their use was justified. Therefore, future research on the juvenile court sentencing process should be conceptualized to reflect the distinctive role of social background variables. Statistical data, notes, and 27 references are included.