In identifying serious juvenile offenders, the criteria used were the type of property offense and the extent of the monetary loss, as well as prior record. The performance of 4,032 serious offenders in restitution projects was examined. Recidivism rates for serious offenders in experimental and control groups were compared in five intensive evaluation sites: Oklahoma County, Okla.; Washington, D.C.; Dane County, Wis.; Ventura County, Calif.; and Ada County, Idaho. Recidivism was measured by official reports of delinquency at 6 and 12 months after the completion of sentences for the experimental and control groups. Serious juvenile offenders completed restitution orders at a rate slightly less than that for all offenders; however, their recidivism rate was greater than that of serious offenders who received traditional dispositions. The recidivism differences were consistent but not statistically significant. The positive view is that serious juvenile offenders can be expected to complete recidivism orders and compensate communities and victims for damage without posing a community threat significantly more serious than offenders receiving traditional dispositions that provide no compensation for damages. Tabular data and 15 references are provided.