The therapeutic index is often used to rank the net effectiveness of drugs and to provide a guide for choosing the best among several drugs. This can be a misleading guide since we show that drugs with the same therapeutic index can have unequal "worth." A substitute ranking measure for the therapeutic index is proposed that would be based on minimizing the losses from the failure to cure plus the losses due to toxicity. This measure would specify a "best" dose over a wide range of conditions, which the therapeutic index does not do. An example is given in which the minimum loss approach produces a ranking of drugs different from that produced by the therapeutic index. In this example the different rankings hold, with minor shifts, over a wide range.