Objectives: Evaluate intra-rater and inter-rater variabilities of the Sunnybrook Facial Grading System (SFGS) and explore potential factors of variability. Study design: Prospective test of hypothesis. Setting: University tertiary referral center. Participants/methods: Facial video recordings of 20 patients with variable degrees of peripheral facial palsy (PFP) were anonymized then presented to 31 independents raters in 2 trials. The raters were senior and junior professionals involved in the management of PFP: ENT specialists, physiotherapists, and speech therapists. The SFGS was used for grading paralysis. Main outcome measure: Intra-rater and inter-rater variabilities were estimated by intraclass correlation coefficient (ICC [95% confidence interval]) for the composite score and the three subscores of the SFGS. Factors of variability studied were: rater professions and rater experience (senior vs junior). Results: For the total population, the intra-rater ICC was 0.915[0.900-0.929] for the composite score considered to represent almost perfect repeatability. Repeatability was important for symmetry at rest (0.694[0.646-0.737]), almost perfect for voluntary movements (0.903[0.886-0.918]), and important to almost perfect for synkinesis (0.810[0.778-0.838]). The inter-rater ICC for the composite score was 0.847[0.755-0.923] indicating important to almost perfect agreement between all raters. Agreement between raters was important to almost perfect for voluntary movements (0.839[0.746-0.919]), but moderate to important for symmetry at rest (0.553[0.408-0.730]) and weak to important for synkinesis (0.476[0.333-0.666]). Some variability was found between raters groups; however, repeatability and agreement were good for all raters. Conclusions: The SFGS is a reproducible scale. It can be used with good reproducibility by both novices and experts, and by all professionals involved in the management of PFP.