• There are inherent methodologic problems in investigating senile dementia of the Alzheimer type (SDAT) but this study indicated that, with regard to behavioral test measurement, some of the obvious problems are partially mitigated. First, these data showed that the progressive decline seen in longitudinal SDAT studies is of the same type and magnitude seen in cross-sectional comparisons. Thus, comparing SDAT groups differing in severity provides for the same observations as following up subjects with mild SDAT over time. This study also demonstrates that with certain kinds of tests, selective subject attrition may not be any greater in longitudinal investigations than found in normal aging research. Subjects who dropped out of the longitudinal study were not that different from those who remained in, at least in terms of initial test performance on challenging tests. In this study less than 5% of the subjects with SDAT originally classified as being in the mild stage of the disease remained in this stage after 63/4 years of investigation. This inexorable downward path of the subject with SDAT is what appears paramount. Given the similarity of staging in longitudinal and cross-sectional studies, prior test performance levels in the latter studies can be ascertained even if with less accuracy than in longitudinal studies.