This prospective study compared the test-retest reliability of thirteen variables calculated from the pendulum test in able-bodied children to those of children diagnosed with cerebral palsy. Ten healthy children and 10 children with a primary diagnosis of cerebral palsy (CP) (mean age 13 years) participated in the study. Data were collected using a three-dimensional motion analysis system on two separate occasions 73+/-28 days apart. The between day reliability ICC scores of all variables were moderate to very high (0.60-0.98) for children with CP and high to very high (0.71-0.98) for able-bodied children. The children with CP demonstrated slower maximum angular velocity compared to the able-bodied children (202 degrees /s versus 293 degrees /s, p<0.01). The time to maximum angular velocity occurred sooner for children with CP compared to able-bodied children (0.22s versus 0.34s, p<0.001). For some children with CP, the knee motions demonstrated were not oscillations of decreasing magnitude. Therefore the integrals of knee motion in each plane were calculated. For both groups of subjects the largest integrals of motion were in the sagittal plane (knee flexion/extension). The able-bodied subject's integrals were twice as large compared to subjects diagnosed with CP (p<0.01). High test-retest reliability of the variables suggests that the pendulum test provides an objective and reliable method to assess quadriceps spasticity in children with cerebral palsy.