The Y448H mutation in NS5B has been selected by GS-9190 as well as several benzothiadiazine hepatitis C virus (HCV) polymerase inhibitors in vitro and in vivo. However, the level and the evolution kinetics of this resistance mutation prior to and during treatment are poorly understood. In this study, we developed an allele-specific real-time PCR (AS-PCR) assay capable of detecting Y448H when it was present at a level down to 0.5% within an HCV population of genotype 1a or 1b. No Y448H mutation was detected above the assay cutoff of 0.5% in genotype 1b-infected Con-1 replicons prior to in vitro treatment. However, the proportion of replicons with the Y448H mutation rapidly increased in a dose-dependent manner upon treatment with GS-9190. After 3 days of treatment, 1.2%, 6.8%, and >50% of the replicon population expressed Y448H with the use of GS-9190 at 1, 10, and 20 times its 50% effective concentration, respectively. In addition, plasma from 65 treatment-naïve HCV-infected patients (42 and 23 with genotype 1a and 1b, respectively) was tested for the presence of Y448H by AS-PCR and population sequencing. As expected, all patient samples were wild type at NS5B Y448 by population sequencing. AS-PCR results were obtained for 62/65 samples tested, with low levels of Y448H ranging from 0.5% to 3.0% detected in 5/62 (8%) treatment-naïve patient samples. These findings suggest the need for combination therapy with HCV-specific inhibitors to avoid viral rebound of preexisting mutant HCV.