Background Data on persistence of NS5A resistance-associated substitutions (RASs) may have implications for resistance testing approaches and selection of initial and retreatment strategies. Methods Long-term persistence of NS5A RASs in HCV genotype (GT) 1 infected subjects ( n=76) who did not achieve sustained virological response after receiving ledipasvir (LDV) without sofosbuvir (SOF) and were subsequently enrolled in an ongoing 3-year follow-up registry study was investigated by population or deep sequencing. Results Of the 76 subjects enrolled, 67 and 9 subjects had GT1a and GT1b infection, respectively. At pretreatment, NS5A RASs were detected in 14% of subjects (11/76) by population sequencing, with three subjects having >1 RAS. All RASs that were detected at pretreatment persisted and were observed at the 96 week visit in the follow-up study (FU96). For the remaining subjects with no detectable RASs at pretreatment, RASs were detected in 98% (63/64) of subjects at virological failure in the parent study and persisted at detectable levels through FU96 in 86% of subjects by deep sequencing (1% cutoff). However, a decline in the quasispecies frequency of most RASs and the number of RASs per subject was observed over time. Phenotypic analysis demonstrated that the majority of NS5A RASs confer similar levels of resistance to LDV and daclatasvir. Conclusions The majority of NS5A RASs can persist at detectable levels for >96 weeks post-treatment in subjects who failed treatment with regimens containing an NS5A inhibitor without SOF, suggesting relatively high fitness of NS5A RASs even in the absence of drug pressure.