Co-offending has a major impact on the arithmetic of crime rates and the burdens on the justice system. This paper studies co-offending by single year of age using data that comprise 750,000 negative police contacts (those charged, chargeable and suspected in criminal offenses) in a largely metropolitan dataset from British Columbia, Canada, 2002-06. We find that shifts in co-offending rates within teenage years are extremely rapid and highly sensitive to sample age ranges, such that a single co-offending rate for all teenagers is misleading. Co-offending opens a range of policy options and issues concerning the presence of youth hangouts and offender convergence settings that can assist the search for suitable co-offenders. Table, figures, and references (Published Abstracts)