Androgen deficiency in women is increasingly recognized as a new clinical syndrome and has raised our awareness of the importance of accurate and well-validated measurements of serum free testosterone (T) concentrations in women. Therefore, we compared serum free T levels measured by equilibrium dialysis to those measured by a direct RIA (analog method) and to those calculated from the law of mass action (requires the measurement of total T and SHBG). We also calculated the free androgen index, 100 x T/SHBG, as a simple index known to correlate with free T. Subjects were 147 women with variable androgen and estrogen statuses. All were studied three times in 1 month and included women 1) with regular menses (estrogen positive, T positive), 2) more than 50 yr old and not receiving estrogen (estrogen negative, T positive), 3) receiving estrogen (estrogen positive, T negative), and 4) with severe androgen deficiency secondary to hypopituitarism (estrogen negative, T negative). Calculated values for free T using the laws of mass action correlated well with those obtained from equilibrium dialysis (r = 0.99; P < 0.0001). However, the agreement depended strongly on the specific assays used for total T and SHBG. In contrast, the direct RIA method had unacceptably high systematic bias and random variability and did not correlate as well with equilibrium dialysis values (r = 0.81; P < 0.0001). In addition, the lower limit of detection was higher for the direct RIA than for equilibrium dialysis or calculated free T. Free androgen index correlates well with free T by equilibrium dialysis (r = 0.93; P < 0.0001), but is a unitless number without reference to the physical reality of free T. We conclude that the mass action equation and equilibrium dialysis are the preferred methods for use in diagnosing androgen deficiency in women.