Phosphomannomutase deficiency (PMM2-CDG) is the most frequent congenital disorder of glycosylation. The cerebellum is nearly always affected in PMM2-CDG patients, a cerebellar atrophy progression is observed, and cerebellar dysfunction is their main daily functional limitation. Different therapeutic agents are under development, and clinical evaluation of drug candidates will require a standardized score of cerebellar dysfunction. We aim to assess the validity of the International Cooperative Ataxia Rating Scale (ICARS) in children and adolescents with genetically confirmed PMM2-CDG deficiency. We compare ICARS results with the Nijmegen Pediatric CDG Rating Scale (NPCRS), neuroimaging, intelligence quotient (IQ) and molecular data. Our observational study included 13 PMM2-CDG patients and 21 control subjects. Ethical permissions and informed consents were obtained. Three independent child neurologists rated PMM2-CDG patients and control subjects using the ICARS. A single clinician administered the NPCRS. All patients underwent brain MRI, and the relative diameter of the midsagittal vermis was measured. Psychometric evaluations were available in six patients. The Mann-Whitney U test was used to compare ICARS between patients and controls. To evaluate inter-observer agreement in patients' ICARS ratings, intraclass correlation coefficients (ICC) were calculated. ICARS internal consistency was evaluated using Cronbach's alpha. Spearman's rank correlation coefficient test was used to correlate ICARS with NPCRS, midsagittal vermis relative diameter and IQ. ICARS and ICARS subscores differed between patients and controls (p < 0.001). Interobserver agreement of ICARS was "almost perfect" (ICC = 0.99), with a "good" internal reliability (Cronbach's alpha = 0.72). ICARS was significantly correlated with the total NPCRS score (r 0.90, p < 0.001). However, there was no agreement regarding categories of severity. Regarding neuroimaging, inverse correlations between ICARS and midsagittal vermis relative diameter (r −0.85, p = 0.003) and IQ (r −0.94, p = 0.005) were found. Patients bearing p.E93A, p.C241S or p.R162W mutations presented a milder phenotype. ICARS is a reliable instrument for assessment of PMM2-CDG patients, without significant inter-rater variability. Despite our limited sample size, the results show a good correlation between functional cerebellar assessment, IQ and neuroimagingFor the first a correlation between ICARS, neuroimaging and IQ in PMM2-CDG patients has been demonstrated.