The aim of this study was to determine in patients with HCV genotype 2 or 3 the performance at week 4 of two assays with different sensitivities for HCV RNA detection, for the prediction of SVR and stratification for treatment duration (14 and 24 weeks). Recruitment was from two trials comparing 14 and 24 weeks treatment to patients with rapid virological response (RVR) (n = 550). RVR was originally defined as HCV RNA <50 IU/ml at week 4. Patients with an available frozen plasma sample drawn at week 4 and with follow-up data week 24 post-treatment were included (n = 429). HCV-RNA was prospectively measured with COBAS Amplicor V2, Roche (CA) (lower detection limit 50 IU/ml) and retrospectively assessed with VERSANT HCV-RNA Qualitative Assay, Siemens (TMA) (lower limit detection 10 IU/ml). Genotype 3 was present in 80% and genotype 2 in 20%. A SVR was achieved in 82%. At week 4 HCV-RNA was undetectable in 74.8% and 63% of serum samples tested with CA and TMA, respectively. CA undetectable/TMA positive was observed in 61/341 (18%) of the samples. In genotype 3 patients a relapse was seen in 9% of the patients with both CA and TMA undetectable and in 25% of the patients who were CA undetectable/TMA positive (p = 0.006). In patients allocated to 14 weeks treatment a relapse was observed in 11% of TMA undetectable patients and 26% of TMA positive (p = 0.031). In genotype 2 patients treated for 14 weeks relapse was observed in 6% of the patients with both CA and TMA undetectable week 4. Assays with high sensitivity for HCV RNA identifies patients at week 4 with high risk of virological relapse. We recommend that patients with genotype 3 and detectable HCV RNA at levels below 50 IU/ml do not receive truncated therapy with pegIFN and ribavirin.