The unbound fraction of plasma testosterone (T) can freely enter tissues, whereas the bioavailability of the albumin-bound T is controversial. A clinical observation in hirsute women receiving spironolactone suggested an experimental paradigm to test the effect of albumin binding on T bioavailability. We found an increase in the non-T-estrogen-binding globulin-bound fraction of plasma T in women from 24.1 +/- 3.9% to 42.0 +/- 8.1% (+/-SEM) while they received spironolactone. Computer modeling indicated that the absolute increase in the albumin-bound T concentration would be about 22.4-fold greater than that in the unbound T concentration (the ratio of albumin-bound to free T remaining virtually constant) because of the binding of T to albumin. We reasoned that the addition of graded amounts of spironolactone and its metabolites to plasma would provide a means to increase the albumin-bound T concentration appreciably. We evaluated the biological effects of this perturbation of T transport by spironolactone and its metabolites in a bioassay system using the Oldendorf technique. Bioavailable T increased proportionately with increments in free and albumin-bound T (r = 0.85; P less than 0.01). A major portion of the albumin-bound T (i.e. 55%) entered tissues under all conditions; the amount that was bioavailable vastly exceeded the amount of T that was unbound in the injected samples. An index of the amount of bioavailable T can be determined using the ammonium sulfate precipitation technique, as the percentage of non-T-estrogen-binding globulin-bound T in vitro correlated well with T bioavailability in vitro (r = 0.86; P less than 0.01). These studies support the conclusion that albumin-bound T is biologically important.