This study investigates whether children and adolescents in treatment systematically vary in their speed of improvement and if speed of improvement is predictable from baseline information. Two studies were conducted using the Fort Bragg Evaluation sample of 984 treated children. In the first study, a cluster analysis of cases sorted the Child Behavior Checklist timelines over 18 months into four patterns of outcome. Three categories showed that youths improved at different rates while one group showed no improvement. This finding was replicated using an independent sample from another study. In a second study, none of 43 child and family characteristics known at intake distinguished early improvers from later improvers of equal severity. Claims to identify how much treatment a child will need should be viewed skeptically until valid prediction has been demonstrated; for now, concurrent monitoring offers an alternative for determining when further treatment yields diminishing returns.