Results indicate two factors that may explain the discrepancy between the observed and predicted reconviction rates in 2006-2007: a higher than expected rate of reconviction among juveniles handled by Youth Justice Conference (YJC), and the absence in the Group Risk Assessment Model (GRAM) of any control for the number of prior police cautions received by a juvenile offender. The higher than expected rate of reconviction among juveniles handled by the YJC in 2006-2007 may be attributable to changes in policing practices. For juveniles who completed an outcome plan for YJC for the index appearance, the seriousness of the first re-offense decreased over time. Further, the re-offense types which increased in prevalence over time appeared to be related to changes in policing practice, with reconvictions that involved offenses against justice procedures increasing over time among offenders handled by YJC at their index appearance. Reconviction rates were investigated on cohorts of juvenile offenders given non-custodial sanctions (for 2004-2005 n=4,225; for 2006-2007 n=4,368). Tables, notes, and references