Aim: To determine (1) the reproducibility of the Eating and Drinking Ability Classification System (EDACS); (2) EDACS classification distribution in a population-based cohort with cerebral palsy (CP); and (3) the relationships between the EDACS and clinical mealtime assessment, other classifications, and health outcomes. Method: This was a cross-sectional population-based cohort study of 170 children with CP at 3 years to 5 years (mean 57.6mo, standard deviation [SD] 8.3mo; 105 males, n=65 females). Functional abilities were representative of a population sample (Gross Motor Function Classification System level I=74, II=34, III=21, IV=18, V=23). The EDACS was the primary classification of mealtime function. The Dysphagia Disorders Survey was the clinical mealtime assessment. Gross motor function was classified using the Gross Motor Function Classification System. Results: EDACS classification had 88.3% intrarater agreement (κ=0.84, intraclass correlation coefficient=0.95; p<0.001) and 51.7% interrater agreement (κ=0.36, intraclass correlation coefficient=0.79; p<0.001). In total, 56.5% of children were classified as EDACS level I. There was a strong stepwise relationship between the Dysphagia Disorders Survey and EDACS (r=0.96, p<0.001). Parental stress (odds ratio=1.3, p=0.05) and feeding tubes (odds ratio=6.4, p<0.001) were significantly related to more limited function on the EDACS. Interpretation: The EDACS presents a viable adjunct to clinical assessment of feeding skills in children with CP for use in surveillance trials and clinical practice. A rating addendum would be a useful contribution to the tool to enhance reproducibility.