The molecular diagnosis of type 1 FacioScapuloHumeral Dystrophy (FSHD1) relies on the detection of a shortened D4Z4 array at the 4q35 locus while until recently, the diagnosis of FSHD2 relied on the absence of a shortened D4Z4 allele in clinically affected patients. The vast majority of FSHD2 patients carry a heterozygous variant in the SMCHD1 gene. In addition, a decreased in D4Z4 DNA methylation is consistently associated with FSHD1 and FSHD2. In molecular genetic diagnostics, predicting the pathogenicity of SMCHD1 variants remains challenging, as many are classified as variants of unknown significance or likely pathogenic. To refine the diagnosis of FSHD2, define 4q-associated molecular features and validate the pathogenicity of SMCHD1 variants, we explored a cohort of 54 FSHD2 patients carrying a variant in SMCHD1 or hemizygosity of the 18p32 locus encompassing the gene. Genetic and epigenetic analyses together with a clinical description of patients were combined to confirm the pathogenicity of new SMCHD1 variants and previously reported ones initially classified as likely pathogenic. We defined a threshold of 40% of methylation at the D4Z4 DR1 site as associated with SMCHD1 pathogenic variants. We also showed that the number of D4Z4 units on the shortest 4qA allele ranges from 11 up to 35 units in patients clinically affected with FSHD2. Using prediction tools, our study further highlighted the difficulty in interpretating the impact of pathogenic variants on the severity of the disease. Our study further emphasizes the complex relationship between D4Z4 methylation, SMCHD1 variants, and disease penetrance in FSHD.