There are many analogies between antineoplastic therapy and antiviral therapy. For each there may be sanctuary sites in which the drug is ineffective because of decreased accumulation of the active form of the drug or increased competition by naturally occurring inhibitors. These sanctuaries may be restricted to anatomic or biochemical subsets of the population. A knowledge of these sanctuaries is essential to an understanding of the failure of therapy and for the design of more effective treatments. Eradication of these sanctuary sites may be important because they may be responsible for the viral replication or tumor cell division that continues to generate the diversity that drives clonal evolution. Ultimately, diversity as a consequence of the accumulation of mutations results in the selection of resistant viral or tumor cell variants and the failure of drug therapy. Maximizing therapy in an attempt to diminish the rate of generation of this diversity may result in better clinical outcomes, including a delay in the generation of variants with genetic drug resistance.