Hormone replacement therapy (HRT) has been proposed for the prevention and treatment of many chronic conditions, ranging from osteoporosis, heart disease, urinary incontinence, and Alzheimer's disease. With the exception of osteoporosis, however, many of the suggested benefits remain controversial. Part of the controversy stems from the relative absence of randomized controlled trials, particularly those enrolling sufficient numbers of elderly women. We propose that another factor may also contribute, one that has been overlooked - failure to consider the variable endogenous estrogen status of elderly women. Highly variable levels of estrogens are present in nearly all postmenopausal women, even at advanced ages. Similar to other endocrine systems, estrogen deficiency and the need for its replacement are, therefore, likely to be relative rather than absolute. Recent studies indicate that elderly women who are less able to compensate for declining ovarian 17beta-estradiol production by adipose synthesis of estrone (E1) may be at greater risk for certain chronic conditions associated with relative estrogen deficiency. Because many markers of estrogen deficiency exhibit overlap between risk groups, their clinical usefulness as predictors of frailty, disability, and response to HRT has been limited. Future studies will need to focus not only on the use of highly variable circulating serum estrogen levels but also on markers of overall estrogenic effects at the level of individual target tissues (i.e., markers of bone turnover, karyopyknotic index on a vaginal wall smear). We propose that a clinical approach that takes into consideration the remarkable heterogeneity (physiological as well as psychological) of elderly women will enable us to approach the decision about HRT in a more individualized and possibly better targeted fashion.