To the Editor: Dr. Smith et al. write to clarify a point that they believe could be misconstrued regarding my statement on the relationship between cognitive—more specifically, visuospatial—function and testosterone levels, i.e., that a curvilinear relationship exists. Actually, we do not disagree on this point, because I indicate that the interpretation of, and comparisons between, results of the different studies on this subject is hampered by the multiple methodological differences that include variables such as handedness. Janowsky et al., 1994, have also noted that "it may be that testosterone has a curvilinear relationship to spatial cognitive performance" (p. 330). Recognizing, however, that this relationship is the subject of debate (1), I chose to write, "it appears that there exists a curvilinear relationship between testosterone level and performance on tests of spatial cognition" (Sternbach, 1998, p. 1314), rather than make a more definitive statement of such a relationship. Undoubtedly, the relationship between testosterone levels in both men and women and cognitive function is a complex one that is influenced by multiple other hormonal and nonhormonal variables.