How to manage patients who have failed one or more intensive antiretroviral regimens is one of the most vexing questions in the current treatment of HIV. An increasingly common scenario involves a patient who may have initially responded favorably to antiretroviral therapy, but whose viral load is now increasing. Such a patient may have been exposed to numerous antiretroviral drugs in a succession of combination regimens. In these scenarios, consultants are often asked to help design salvage regimens. When a patient's antiretroviral therapy is identified as "failing," the natural impulse is to seek another regimen that will succeed. The desire to fully suppress viral replication, however, must be tempered by a full appreciation of the implications of changing therapy. The clinician should consider two factors. First, the overall success rates of salvage regimens are significantly lower than the success rates of initial therapies, probably for two reasons: (a) cross-resistance between agents used in the salvage regimen and agents used in previous failing regimens and (b) reduced adherence to salvage regimens caused by increased complexity, pill burden, and toxicity. Second, if a salvage regimen is not fully suppressive or is not taken correctly, it may lead to the selection of additional resistance mutations, making the patient's virus even more difficult to treat in the future. Every change in therapy, therefore, carries significant risks. Such changes should generally be undertaken with reluctance and only when the anticipated benefits clearly outweigh these risks. Unfortunately, no clear definition of treatment failure is available to guide clinicians in changing therapy. Table 1 lists some of the many definitions of treatment failure currently in use among HIV clinicians. Depending on the definitions chosen (and the sensitivity of the viral-load assay used), many patients are virologic failures but remain immunologic and clinical successes. Although there is good …