The role of prognostic factors in optimizing treatment for breast cancer patients has clearly changed with the trend toward general use of adjuvant therapy. Nevertheless, there are at least three situations in which prognostic factors could be helpful. The first is to identify patients whose prognosis is so good that adjuvant therapy after local surgery would not be cost-beneficial. The second is to identify patients whose prognosis is so poor that a more aggressive adjuvant approach would be warranted. The third is to identify patients likely to be responsive or resistant to particular forms of therapy. Here we will discuss all of these situations, together with the appropriate cutpoint analyses and validation of individual prognostic factors and their combination into prognostic indexes which may make the discrimination of patient subsets more reliable.