BACKGROUND Low-level viremia (LLV) is measurable, with enhanced assays, in many subjects with HIV RNA levels <50 copies per milliliter. The clinical consequences of LLV are unknown. METHODS In a prospective study in HIV-1-infected adults, HIV RNA levels were determined with an ultrasensitive test (3 copies/mL) based on a real time polymerase chain reaction. The primary end point was to evaluate LLV prediction of virological failure, defined as a confirmed plasma HIV RNA level >50 copies per milliliter. RESULTS One thousand two hundred fourteen patients were followed for (mean) 378 days. At baseline, 71.5% were <3 copies per milliliter below the limit of detection (BLD). The risk of failing highly active antiretroviral therapy in the following 4 months for patients BLD was 0.4% compared with a 3.2% risk for those with LLV (P < 0.0001; odds ratio: 7.52). There was a significant (P < 0.0001) linear relationship between the HIV RNA and the risk of virologic failure. LLV receiver operating curve analysis showed an area under the curve of 0.76 (95% confidence interval: 0.68 to 0.84) that significantly (P < 0.0001) predicted the risk of failure. The risk of an unconfirmed viral blip was higher in patients with LLV (3.9%) than in those BLD (1.1%) (P < 0.0001; odds ratio: 3.56). Longer exposure to antiretrovirals, current use of nonnucleoside reverse transcriptase inhibitors, longer time BLD, and current HIV RNA <3 copies per milliliter were independent predictors of a positive outcome. CONCLUSIONS Viral replication may be the cause of LLV, at least in some patients. A LLV >3 copies per milliliter is linked to a significant increment of risk of virological failure leading to drug resistance. Patients with measurable LLV should be managed to better evaluate, over time, the risk of failure and to limit its consequences.