Aim This study aimed to investigate the role of resistance‐associated substitutions (RASs) to direct‐acting‐antivirals (DAAs) in HCV genotype 3 (GT3). Methods Within the Italian VIRONET‐C network, a total of 539 GT3‐infected patients (417 DAA‐naïve and 135 DAA‐failures, of them, 13 at both baseline and failure) were analysed. Sanger sequencing of NS3/NS5A/NS5B was performed following home‐made protocols. Results The majority of patients were male (79.4%), 91.4% were injection drug users, 49.3% were cirrhotic and 13.9% were HIV co‐infected. Phylogenetic analysis classified sequences as GT3a‐b‐g‐h (98%‐0.4%‐0.2%‐1.2%) respectively. Overall, 135 patients failed a DAA regimen: sofosbuvir (SOF)/daclatasvir (DCV) or velpatasvir (VEL)±ribavirin (RBV) (N = 91/15) and glecaprevir (G)/pibrentasvir (P) (N = 9). Moreover, 14.8% of patients were treated with suboptimal regimens for GT3: 3D ± RBV (Paritaprevir/r + Ombitasvir+Dasabuvir, N = 15), SOF + Simeprevir (SIM) (N = 1) or SOF/Ledipasvir (LDV) ± RBV (N = 4). RAS prevalence was 15.8% in DAA‐naïve patients. At failure, 81.5% patients showed at least one RAS: 11/25 (44.0%) in NS3, 109/135 (80.7%) in NS5A, 7/111 (6.3%) in NS5B SOF‐failures. In NS5A‐failures, Y93H RAS was the most prevalent (68.5% vs 5.1% DAA‐naïve, P < .001) followed by A30K (12.7% vs 2.8% in DAA‐naïve, P < .001). Analysing baseline samples, a higher prevalence of NS5A‐RASs was observed before treatment in DAA‐failures (5/13, 38.5%) vs DAA‐naïves (61/393, 15.5%, P = .04). Regarding 228 DAA‐naïve patients with an available outcome, 93.9% achieved a SVR. Interestingly, patients with baseline Y93H and/or A30K had SVR rate of 72.2% vs 95.7% for patients without NS5A‐RASs (P = .002). Conclusions In this real‐life GT3 cohort, the majority of failures harboured resistant variants carrying NS5A‐RASs, the most frequent being Y93H. The presence of natural NS5A‐RASs before treatment was associated with failure. Further analyses are needed to confirm this observation, particularly for the new current regimens.