I commend the authors of the recent paper on perceptions of the dying experience of older and seriously ill patients [1] for their efforts to characterize and improve that experience. However, some may be tempted to make inferences that are improper because the study was not controlled. Only the experiences of those patients who did die are described, and we do not know what would have been experienced by a similar set of patients who did not die. Thus, some lay reporters have inferred that the aggressive treatment given in the last 3 days of life was both futile (it was) and inappropriate (it may or may not have been). Here, the heterogeneity of the patient sample makes it difficult to judge appropriateness. Some patients who died (for example, those with congestive failure) may have been indistinguishable from others who survived because of the aggressive treatment and who indeed may have had a gratifying prolongation of life. Although these patients may have expressed a preference for "a treatment plan that focused on comfort," they may, in retrospect, have considered the aggressive treatment well worthwhile. Thus, it would be wrong to draw conclusions about the appropriateness of this "aggressive" treatment without knowing how quality of life in the last 3 days related to the probability of some useful outcome. It would also be helpful to know why aggressive treatment was undertaken in these cases, particularly if such treatment was contrary to the patients' wishes. Perhaps, in what must be an enormous database, the authors have data with which to address these questions. Unfortunately, until we can predict outcomes with more certainty and thereby better advise our patients when treatment is or is not appropriate, we will continue to make mistakes, providing too much or too little treatment.