Objective Cerebral palsy (CP) can be classified as spastic, dyskinetic, ataxic or combined. Correct classification is essential for symptom-targeted treatment. This study aimed to investigate agreement among professionals on the phenotype of children with CP based on standardized videos. Methods In a prospective, observational pilot study, videos of fifteen CP patients (8 boys, mean age 11 ± 5 y) were rated by three pediatric neurologists, three rehabilitation physicians and three movement disorder specialists. They scored the presence and severity of spasticity, ataxia or dyskinesias/dystonia. Inter- and intraobserver agreement were calculated using Cohen's and Fleiss' kappa. Results We found a fair inter-observer (κ = 0.36) and moderate intra-observer agreement (κ = 0.51) for the predominant motor symptom. This only slightly differed within the three groups of specialists (κ = 0.33–0.55). Conclusion A large variability in the phenotyping of CP children was detected, not only between but also within clinicians, calling for a discussing on the operational definitions of spasticity, dystonia and ataxia. In addition, the low agreement found in our study questions the reliability of use of videos to measure intervention outcomes, such as deep brain stimulation in dystonic CP. Future studies should include functional domains to assess the true impact of management options in this highly challenging patient population.