The cause of downbeat nystagmus (DBN) remains unknown in approximately 30% of patients (idiopathic DBN). Here, we hypothesized that: (i) FGF14 (GAA) ≥250 repeat expansions represent a frequent genetic cause of idiopathic DBN syndromes, (ii) are treatable with 4-aminopyridine (4-AP), and (iii) FGF14 (GAA) 200-249 alleles are potentially pathogenic. We conducted a multi-modal cohort study of 170 patients with idiopathic DBN that comprised: in-depth ocular motor, neurological, and disease evolution phenotyping; assessment of 4-AP treatment response, including re-analysis of placebo-controlled video-oculography treatment response data from a previous randomized double-blind 4-AP trial; and genotyping of the FGF14 repeat. Frequency of FGF14 (GAA) ≥250 expansions was 48% (82/170) in the entire idiopathic DBN cohort. Additional cerebellar ocular motor signs were observed in 100% (82/82), cerebellar ataxia in 43% (35/82), and extracerebellar features in 21% (17/82) of (GAA) ≥250 - FGF14 patients. Alleles of 200 to 249 GAA repeats were enriched in patients with DBN (12%; 20/170) compared to controls (0.87%; 19/2,191; OR, 15.20; 95% CI, 7.52-30.80; p =9.876e-14). The phenotype of (GAA) 200-249 - FGF14 patients closely mirrored that of (GAA) ≥250 - FGF14 patients. (GAA) ≥250 - FGF14 and (GAA) 200-249 - FGF14 patients had a significantly greater clinician-reported (80% vs 31%; p =0.0011) and self-reported (59% vs 11%; p =0.0003) response rate to 4-AP treatment compared to (GAA) <200 - FGF14 patients. This included a treatment response with high relevance to everyday living, as exemplified by an improvement of 2 FARS stages in some cases. Placebo-controlled video-oculography data of four (GAA) ≥250 - FGF14 patients previously enrolled in a 4-AP randomized double-blind trial showed a significant decrease in slow phase velocity of DBN with 4-AP, but not placebo. This study shows that FGF14 GAA repeat expansions are a highly frequent genetic cause of DBN syndromes, especially when associated with additional cerebellar features. Moreover, they genetically stratify a subgroup of patients with DBN that appear to be highly responsive to 4-AP, thus paving the way for a "theranostics" approach in DBN syndromes.