Despite the long series of cohort studies performed during the last 20 years, the correlation between serum testosterone and any clinical situation believed to be under androgen control in women has remained elusive. This is likely related to the recent finding that the androgens made locally in large amounts in peripheral tissues from the precursor dehydroepiandrosterone (DHEA) act in the same cells where synthesis takes place and are not released in significant amounts in the circulation, thus making unreliable the measurement of serum testosterone as marker of total androgenic activity. The objective is to determine if serum androgen glucuronides can be replaced by testosterone or another steroid as measure of androgenic activity. Since the glucuronide derivatives of androgens are the obligatory route of elimination of all androgens, these metabolites were measured by liquid chromatography tandem mass spectrometry under basal conditions in 377 healthy postmenopausal women aged 55-65 years as well as in 47 premenopausal women aged 30-35 years while testosterone was assayed by gas chromatography mass spectrometry. No correlation was found between the serum concentration of testosterone and that of androsterone glucuronide (ADT-G) or androstenediol glucuronide (3alpha-diol-G), the androgen metabolites which account for the total pool of androgens. The present data show that measurement of the total pool of androgens reflected by the serum levels of ADT-G and 3alpha-diol-G cannot be replaced by serum testosterone or any other steroid, including DHEA or DHEA sulphate. These findings may have implications for women with androgen deficiency involving osteoporosis, obesity, type 2 diabetes, sexual dysfunction, loss of muscular strength and a series of other clinical situations affecting women's health. Measuring ADT-G and 3alpha-diol-G might identify cases of true androgen deficiency and provide an opportunity to offer appropriate androgen therapy.