Due to the complexity of the genomic region, the relationship between genotype, phenotype and epigenetics in facioscapulohumeral muscular dystrophy remains incompletely understood, and diagnostics are limited in precision. In this study, we applied Oxford Nanopore Technology long-read sequencing to characterize FSHD1 and FSHD2 patients with common and previously confirmed genotypes ( n = 14). This approach enabled the simultaneous determination of allele-specific methylation profiles, haplotypes and repeat lengths. Our results confirm the presence of an asymmetric methylation gradient increasing from the proximal to the distal end of 4q35 repeat arrays with a steeper slope in FSHD1 (2.5-8.1%/repeat unit) compared to the two FSHD2 patients (0.3 and 1.2%/repeat unit) studied. The slope is, therefore, a potential marker to differentiate both types of facioscapulohumeral muscular dystrophy, being less biased by repeat length compared to parameters determined in current bisulfite sequencing-based methylation assays. By determining distal methylation allele-specifically, long-read sequencing overcomes a major limitation of such assays. Distal methylation of pathogenic (range 22-45%) and non-pathogenic repeat arrays (range 77-87%) ( P < 0.0001) is clearly separated and shows a stronger correlation with age-corrected clinical severity ( r = -0.81, 95% CI: -0.48 to -0.94) compared to studies based on bisulfite sequencing. Overall, this strengthens distal methylation as a superior diagnostic marker of facioscapulohumeral muscular dystrophy disease status. Detailed clinical and genetic characterization of a second cohort ( n = 20) comprising patients with complex genotypes ( n = 11) and FSHD2 ( n = 9) was performed using our bisulfite sequencing-based methylation profile analysis, complementary methods and in three cases, long-read sequencing. Based on these results, we highlight underrecognized patient groups at risk of misdiagnosis when methylation analysis is not part of the diagnostic workup: (i) patients carrying a 4qA166 repeat array; (ii) patients with underrepresented population backgrounds; (iii) facioscapulohumeral muscular dystrophy patients carrying 11 repeat units in the absence of global hypomethylation; and (iv) FSHD2 patients without a detectable causative variant in known epigenetic suppressor genes ( SMCHD1, DNMT3B and LRIF1). We hypothesize that unknown epigenetic modifiers explain the penetrance of repeat arrays with 8-10 repeat units in the European and North American populations, the pathogenicity of some repeat arrays with 11 repeat units and the hypomethylation of the entire 4q35 repeat array in some FSHD2 patients without a detected variant in known epigenetic suppressor genes. From our results, recommendations for standard diagnostics are derived. Finally, we hypothesize a preliminary model linking DNA methylation, known genetic parameters and disease severity.