AIM The goal of this study was to contribute evidence towards the effectiveness of task-oriented training with and without restriction of trunk movement (trunk restraint) on the quality of upper limb movement in children with cerebral palsy (CP).METHOD We used a prospective, single-subject research design in 12 children (three males, nine females; aged 6-11y; median 9y) with di-, hemi-, or quadriplegia. Movements of the most affected arm were assessed five times: three times before training, immediately after training, and 3 months after training. The main outcome measures were the Melbourne Assessment of Unilateral Upper Limb Function (Melbourne) and upper limb movement kinematics during a functional reaching task. Children were randomly allocated to one of two groups: task-oriented training with or without trunk restraint. Treatment consisted of three 1-hour sessions per week for 5 weeks (total training duration 15h). Treatment effects were determined using single-subject research design analysis -regression through baseline data and standard mean differences.RESULTS Although the Melbourne scores were largely unchanged after training, some children in each group improved arm trajectory smoothness (effect size 0.55-1.87), and most children improved elbow extension range (effect size 0.55-4.79). However, more children in the trunk restraint group than in the no restraint group demonstrated reduced trunk displacement (effect size 0.94-2.25) and longer-term improvements in elbow extension and trunk use. Among the group who underwent training without trunk restraint, trunk displacement was unchanged or increased, and fewer carry-over effects were apparent at follow-up.INTERPRETATION This proof-of-principle study showed that greater improvement in the quality of upper limb movement in children with CP, including less compensatory trunk use and better carry-over effects, was achieved by training with trunk restraint.