The present report describes a RIA for 3',5'-diiodothyronine (T2) that can be performed on unextracted serum and which has a lower limit of detectability of 2 ng/dl. Cross-reactivity with other iodothyronines was negligible, except for rT3 which began to demonstrate cross-reactivity when rT3 levels were elevated to 180 ng/dl. Employing this RIA for T2, we have determined that 83 healthy individuals had a mean (+/-SE) serum T2 concentration of 5.0 +/- 0.3 ng/dl, thyrotoxic subjects (n = 12) had a mean T2 level that was elevated to 10.8 +/- 0.8 ng/dl, and each of 6 hypothyroid subjects had undetectable (less than 2 ng/dl) concentrations. Athyreotic patients (n = 8), receiving 0.4 mg T4 daily, had serum T2 concentrations of 15.0 +/- 3.0 ng/dl. Fasting in obese subjects was associated with an increase in serum T2 to 6.9 +/- 0.6 ng/dl from a basal level of 4.4 +/- 0.4 ng/dl in the fed state (P less than 0.01). Despite the fact that rT3 levels may be elevated in amniotic fluid and that rT3 is expected to represent the major source from which extrathyroidal T2 arises, T2 levels were low in amniotic fluid, being undetectable (less than 2 ng/dl) in 9 of 19 samples; the mean (+/-SE) T2 concentration in the 10 detectable samples was 5.4 +/- 1 ng/dl. These data indicate T2 is a normal component of serum and that the majority of serum T2 is probably derived from peripheral conversion. Furthermore, these observations suggest that situations associated with elevated rT3 levels (e.g. thyrotoxicosis and fasting) may also have increased T2 values.