The criteria for evaluating response to treatment for patients with adult acute leukemia are objective, quantitative, and effective. The first objective of treatment is to achieve a complete clinical and hematological remission, since in the absence of such response the disease has a rapidly progressive and inevitably fatal outcome. Logistic regression analysis and the generation of models for predicting the probability of response for each patient have proven to be extremely useful techniques for evaluating new candidates for front-line treatment in acute leukemia. In prospective tests, such modeling procedures have proven to predict accurately for response. It is evident that this methodology will be highly effective for selecting treatment for individual patients and for detecting improved treatments for patients who respond poorly to conventional treatment. For patients who achieve complete remission, the duration of that remission can also be evaluated by logistic regression analysis. Variables which predict for duration of remission have proven to be different from those variables which predict for probability of response. In addition, variables which predict for prolonged remission--that is, beyond 2 and 3 years--have proven to be different from variables which predict for relapse in the first year of remission. Therefore, the models which estimate risk of relapse per unit of time at risk should prove most valuable for evaluating not only the choice of treatments to be used during remission, but the duration of treatment and the strategy of treatment, that is, whether intensive treatment or low-dose intermittent treatment should be chosen. Finally, the demonstration that a significant fraction of the patients in complete remission will have long term disease-free and treatment-free survival makes the identification of such patients even more important. Evaluation of innovative higher-risk strategies for treatment of the patient in remission should be offered primarily to patients who have a low probability of having already been cured with conventional treatment. Thus, the rapid evolution of effective treatment for adult acute leukemia over the last 15 years has necessitated the development of more precise and more effective methods for evaluating the effects of those treatments on patients with this disease.