In PNAS, Hsu et al. (1) observed an early transient increase (here, termed a blip) in hepatitis C virus (HCV) viral load in 92 out of 145 Taiwanese patients with chronic hepatitis C treated with pegylated IFN-α (IFN) and ribavirin and found that these blips were significantly associated with the outcome of therapy [i.e., sustained virologic response (SVR)]. Further, they sought to explain the origin and time of these blips using a version of a well-studied mathematical model of HCV infection and treatment (2) in a nonstandard way. Although these blips are interesting, several concerns need to be raised before further interpretation of the findings of Hsu et al. (1) is made.During chronic HCV infection, the serum level of HCV RNA can vary (up to 0.5-log) on time scales of weeks to months (3). In the absence of large studies with time scales of hours, it is difficult to verify if all the blips observed were generated by IFN/ribavirin therapy. Nonetheless, Hsu et al. (1) showed that among patients who had blips, a significantly higher proportion of patients with SVR had early blips, suggesting that a large proportion of the blips were related to therapy. Because patients infected with HCV genotype 1 respond poorly to IFN/ ribavirin compared with those infected with HCV genotype 2, the distribution of the blips among HCV genotypes would be relevant.A basic assumption in modeling HCV kinetics during treatment is that viral load is constant before therapy, implying that the rate of viral production equals the rate of virion clearance in the absence of therapy (2). If one makes this assumption, one can then show that therapy, which decreases viral production, will lead to a decrease in viral load and cannot reproduce viral blips. In the three-parameter model solution used by Hsu et al. (1) [i.e., V(t = time) = (K 1 ′t + K 2 ′)e −ct )], imposing pretreatment steady state implies K 2 ′ = V 0 , the baseline viral load, and K 1 ′ = (1 − ε)cV 0 , where ε is the effectiveness of treatment in blocking virion production and c is the virion clearance rate (2). Hsu et al. (1) did not assume viral steady state before therapy and left K 1 ′ and K 2 ′ as free parameters without justification. Moreover, the interpretation of the blips is hampered by a number of questionable assumptions made in deriving the three-parameter model solution, such as therapy being 100% effective in blocking new infections and δ, the rate of infected cell death, set equal to c, despite prior studies showing c is ∼50-fold larger than δ (4).Clearly, if some patients were not in steady state and had viral loads increasing before therapy (i.e., viral production higher than viral clearance), then, on therapy initiation, viral loads would continue to increase until the effect of therapy reduces the viral production rate below the viral clearance rate (5). The more effective the therapy, the sooner this should occur, possibly explaining why among patients who had blips, patients with SVR had earlier blips than patients without SVR. This conclusion can be reached...