Common wisdom about juvenile delinquency is largely based on newspaper, radio, and television reports and on hearsay and opinion. Criminologists are often unable to observe juvenile delinquency directly because most juvenile crimes are committed in secrecy and because of ethical reservations about participating in or observing crime. These and other constraints make it necessary to rely mainly on official crime statistics and indirect observations to measure the scale and structure of juvenile delinquency in time and space. In the analysis of juvenile delinquency using criminal statistics, there are at least three sets of data available from three crime control subsystems: police, courts, and prisons. Police criminal statistics record crimes that are known to the police and provide information on suspect and victim age and sex and on damage caused. One way of extending police, court, and prison statistical data longitudinally is to study cohorts. Despite improvements in statistical instruments used, however, offense and offender statistics reflect only part of the real incidence of crime. Many crimes are not detected and, of detected crimes, only a certain percentage come to the attention of formal crime control authorities. Dark field research takes different approaches. In self-reported delinquency surveys, test subjects are asked whether they have ever committed specific crimes or delinquent acts. Victim surveys ask test subjects whether they or anyone they know have been victims of specific crimes. Objections to and methodological problems in dark field research are noted, and the complementary use of criminal statistics and dark field research is discussed. 33 notes