Cluster analysis is applied to substance use and delinquency data coll
ected in a longitudinal study of juvenile detainees to empirically der
ive five groups of youths from information gathered at their initial i
nterviews (time 1): alcohol/marijuana-hashish users, low-level delinqu
ents, alcohol/ marijuana-hashish and cocaine-using nondelinquents, hig
h delinquency, cocaine users, and heavy cocaine-using nondelinquents.
The validity of the typology was supported by a variety of other initi
al interview and follow-up interview alcohol/other drug use and delinq
uency data, and by recidivism information - including data on arrests
during the three-and-a-half years following the date of the youths' fi
rst interviews. Research and policy implications of the findings are d
rawn.